perf: IoT 卡 30M 行分页查询优化（P95 17.9s → <500ms）

- 新增 is_standalone 物化列 + 触发器自动维护（迁移 056） - 并行查询拆分：多店铺 IN 查询拆为 per-shop goroutine 并行 Index Scan - 两阶段延迟 Join：深度分页（page≥50）走覆盖索引 Index Only Scan 取 ID 再回表 - COUNT 缓存：per-shop 并行 COUNT + Redis 30 分钟 TTL - 索引优化：删除有害全局索引、新增 partial composite indexes（迁移 057/058） - ICCID 模糊搜索路径隔离：trigram GIN 索引走独立查询路径 - 慢查询阈值从 100ms 调整为 500ms - 新增 30M 测试数据种子脚本和 benchmark 工具
2026-02-24 16:23:02 +08:00
parent c665f32976
commit f32d32cd36
20 changed files with 2705 additions and 50 deletions
--- a/internal/model/iot_card.go
+++ b/internal/model/iot_card.go
@@ -50,6 +50,7 @@ type IotCard struct {
 	StoppedAt       *time.Time `gorm:"column:stopped_at;comment:停机时间" json:"stopped_at,omitempty"`
 	ResumedAt       *time.Time `gorm:"column:resumed_at;comment:最近复机时间" json:"resumed_at,omitempty"`
 	StopReason      string     `gorm:"column:stop_reason;type:varchar(50);comment:停机原因(traffic_exhausted=流量耗尽,manual=手动停机,arrears=欠费)" json:"stop_reason,omitempty"`
+	IsStandalone    bool       `gorm:"column:is_standalone;type:boolean;default:true;not null;comment:是否为独立卡(未绑定设备) 由触发器自动维护" json:"is_standalone"`
 }

 // TableName 指定表名
--- a/internal/service/iot_card/service.go
+++ b/internal/service/iot_card/service.go
@@ -10,6 +10,7 @@ import (
 	"github.com/break/junhong_cmp_fiber/internal/store/postgres"
 	"github.com/break/junhong_cmp_fiber/pkg/constants"
 	"github.com/break/junhong_cmp_fiber/pkg/errors"
+	"github.com/break/junhong_cmp_fiber/pkg/middleware"
 	"go.uber.org/zap"
 	"gorm.io/gorm"
 )
@@ -125,6 +126,19 @@ func (s *Service) ListStandalone(ctx context.Context, req *dto.ListStandaloneIot
 		filters["series_id"] = *req.SeriesID
 	}

+	// 代理用户注入 subordinate_shop_ids，让 Store 层走并行查询路径
+	// 避免 PG 对 shop_id IN (...) + ORDER BY 选择全表扫描
+	userType := middleware.GetUserTypeFromContext(ctx)
+	if userType == constants.UserTypeAgent {
+		shopID := middleware.GetShopIDFromContext(ctx)
+		if shopID > 0 {
+			subordinateIDs, err := s.shopStore.GetSubordinateShopIDs(ctx, shopID)
+			if err == nil && len(subordinateIDs) > 1 {
+				filters["subordinate_shop_ids"] = subordinateIDs
+			}
+		}
+	}
+
 	cards, total, err := s.iotCardStore.ListStandalone(ctx, opts, filters)
 	if err != nil {
 		return nil, err
--- a/internal/store/postgres/iot_card_store.go
+++ b/internal/store/postgres/iot_card_store.go
@@ -2,14 +2,25 @@ package postgres

 import (
 	"context"
+	"fmt"
+	"hash/fnv"
+	"sort"
+	"sync"
+	"time"

 	"github.com/break/junhong_cmp_fiber/internal/model"
 	"github.com/break/junhong_cmp_fiber/internal/store"
 	"github.com/break/junhong_cmp_fiber/pkg/constants"
+	pkggorm "github.com/break/junhong_cmp_fiber/pkg/gorm"
+	"github.com/break/junhong_cmp_fiber/pkg/logger"
+	"github.com/break/junhong_cmp_fiber/pkg/middleware"
 	"github.com/redis/go-redis/v9"
+	"go.uber.org/zap"
 	"gorm.io/gorm"
 )

+const listCountCacheTTL = 30 * time.Minute
+
 type IotCardStore struct {
 	db    *gorm.DB
 	redis *redis.Client
@@ -102,9 +113,11 @@ func (s *IotCardStore) List(ctx context.Context, opts *store.QueryOptions, filte
 	query := s.db.WithContext(ctx).Model(&model.IotCard{})

 	// 企业用户特殊处理：只能看到授权给自己的卡
+	// 子查询跳过数据权限过滤，权限已由外层查询的 GORM callback 保证
+	skipCtx := pkggorm.SkipDataPermission(ctx)
 	if enterpriseID, ok := filters["authorized_enterprise_id"].(uint); ok && enterpriseID > 0 {
 		query = query.Where("id IN (?)",
-			s.db.Table("tb_enterprise_card_authorization").
+			s.db.WithContext(skipCtx).Table("tb_enterprise_card_authorization").
 				Select("card_id").
 				Where("enterprise_id = ? AND revoked_at IS NULL AND deleted_at IS NULL", enterpriseID))
 	}
@@ -130,7 +143,7 @@ func (s *IotCardStore) List(ctx context.Context, opts *store.QueryOptions, filte
 	}
 	if packageID, ok := filters["package_id"].(uint); ok && packageID > 0 {
 		query = query.Where("id IN (?)",
-			s.db.Table("tb_package_usage").
+			s.db.WithContext(skipCtx).Table("tb_package_usage").
 				Select("iot_card_id").
 				Where("package_id = ? AND deleted_at IS NULL", packageID))
 	}
@@ -151,9 +164,13 @@ func (s *IotCardStore) List(ctx context.Context, opts *store.QueryOptions, filte
 		query = query.Where("series_id = ?", seriesID)
 	}

-	// 统计总数
-	if err := query.Count(&total).Error; err != nil {
-		return nil, 0, err
+	if cachedTotal, ok := s.getCachedCount(ctx, "iot_card", filters); ok {
+		total = cachedTotal
+	} else {
+		if err := query.Count(&total).Error; err != nil {
+			return nil, 0, err
+		}
+		s.cacheCount(ctx, "iot_card", filters, total)
 	}

 	// 分页处理
@@ -181,16 +198,393 @@ func (s *IotCardStore) List(ctx context.Context, opts *store.QueryOptions, filte
 	return cards, total, nil
 }

+// standaloneListColumns 列表查询只选取响应需要的列，避免 SELECT * 的宽行 I/O
+var standaloneListColumns = []string{
+	"id", "iccid", "card_category", "carrier_id", "carrier_type", "carrier_name",
+	"imsi", "msisdn", "batch_no", "supplier", "cost_price", "distribute_price",
+	"status", "shop_id", "activated_at", "activation_status", "real_name_status",
+	"network_status", "data_usage_mb", "current_month_usage_mb", "current_month_start_date",
+	"last_month_total_mb", "last_data_check_at", "last_real_name_check_at",
+	"enable_polling", "series_id", "first_commission_paid", "accumulated_recharge",
+	"created_at", "updated_at",
+}
+
+// ListStandalone 独立卡列表查询入口
+// 路由逻辑：
+//   - ICCID/MSISDN 模糊搜索 → listStandaloneDefault（trigram GIN 索引）
+//   - 多店铺（代理用户） → listStandaloneParallel（per-shop 并行，内含深度分页两阶段优化）
+//   - 深度分页（page >= 50）→ listStandaloneTwoPhase（覆盖索引延迟 Join）
+//   - 其他 → listStandaloneDefault
 func (s *IotCardStore) ListStandalone(ctx context.Context, opts *store.QueryOptions, filters map[string]any) ([]*model.IotCard, int64, error) {
+	if opts == nil {
+		opts = &store.QueryOptions{
+			Page:     1,
+			PageSize: constants.DefaultPageSize,
+		}
+	}
+
+	_, hasICCID := filters["iccid"].(string)
+	_, hasMSISDN := filters["msisdn"].(string)
+	useFuzzySearch := (hasICCID && filters["iccid"] != "") || (hasMSISDN && filters["msisdn"] != "")
+
+	if !useFuzzySearch {
+		if shopIDs, ok := filters["subordinate_shop_ids"].([]uint); ok && len(shopIDs) > 1 {
+			return s.listStandaloneParallel(ctx, opts, filters, shopIDs)
+		}
+	}
+
+	// 非多店铺场景的深度分页走两阶段路径（单店铺或平台用户）
+	if !useFuzzySearch && opts.Page >= 50 {
+		return s.listStandaloneTwoPhase(ctx, opts, filters)
+	}
+
+	return s.listStandaloneDefault(ctx, opts, filters)
+}
+
+// listStandaloneTwoPhase 单路径两阶段延迟 Join
+// 适用于非多店铺场景（平台用户、单店铺）的深度分页（page >= 50）
+// Phase 1: SELECT id 走覆盖索引；Phase 2: WHERE id IN 回表
+func (s *IotCardStore) listStandaloneTwoPhase(ctx context.Context, opts *store.QueryOptions, filters map[string]any) ([]*model.IotCard, int64, error) {
+	var total int64
+
+	query := s.db.WithContext(ctx).Model(&model.IotCard{}).
+		Where("is_standalone = true")
+	query = s.applyStandaloneFilters(ctx, query, filters)
+
+	if cachedTotal, ok := s.getCachedCount(ctx, "standalone", filters); ok {
+		total = cachedTotal
+	} else {
+		if err := query.Count(&total).Error; err != nil {
+			return nil, 0, err
+		}
+		s.cacheCount(ctx, "standalone", filters, total)
+	}
+
+	offset := (opts.Page - 1) * opts.PageSize
+
+	// Phase 1: 仅取 ID（覆盖索引 Index Only Scan）
+	var ids []uint
+	if err := query.Select("id").
+		Order("created_at DESC, id").
+		Offset(offset).Limit(opts.PageSize).
+		Pluck("id", &ids).Error; err != nil {
+		return nil, 0, err
+	}
+
+	if len(ids) == 0 {
+		return []*model.IotCard{}, total, nil
+	}
+
+	// Phase 2: 用 ID 精确回表
+	var cards []*model.IotCard
+	if err := s.db.WithContext(ctx).Model(&model.IotCard{}).
+		Select(standaloneListColumns).
+		Where("id IN ?", ids).
+		Find(&cards).Error; err != nil {
+		return nil, 0, err
+	}
+
+	idOrder := make(map[uint]int, len(ids))
+	for i, id := range ids {
+		idOrder[id] = i
+	}
+	sort.Slice(cards, func(i, j int) bool {
+		return idOrder[cards[i].ID] < idOrder[cards[j].ID]
+	})
+
+	logger.GetAppLogger().Debug("两阶段查询完成",
+		zap.Int("page", opts.Page),
+		zap.Int("phase2_fetched", len(cards)),
+		zap.Int64("total_count", total),
+	)
+
+	return cards, total, nil
+}
+
+// listStandaloneDefault 原始查询路径
+// 适用于平台/超管用户（无 shop_id 过滤）或单店铺场景
+func (s *IotCardStore) listStandaloneDefault(ctx context.Context, opts *store.QueryOptions, filters map[string]any) ([]*model.IotCard, int64, error) {
 	var cards []*model.IotCard
 	var total int64

-	query := s.db.WithContext(ctx).Model(&model.IotCard{})
+	query := s.db.WithContext(ctx).Model(&model.IotCard{}).
+		Where("is_standalone = true")
+	query = s.applyStandaloneFilters(ctx, query, filters)

-	query = query.Where("id NOT IN (?)",
-		s.db.Model(&model.DeviceSimBinding{}).
-			Select("iot_card_id").
-			Where("bind_status = ?", 1))
+	if cachedTotal, ok := s.getCachedCount(ctx, "standalone", filters); ok {
+		total = cachedTotal
+	} else {
+		if err := query.Count(&total).Error; err != nil {
+			return nil, 0, err
+		}
+		s.cacheCount(ctx, "standalone", filters, total)
+	}
+
+	offset := (opts.Page - 1) * opts.PageSize
+
+	if opts.OrderBy != "" {
+		query = query.Order(opts.OrderBy)
+	} else {
+		query = query.Order("created_at DESC")
+	}
+
+	if err := query.Select(standaloneListColumns).Offset(offset).Limit(opts.PageSize).Find(&cards).Error; err != nil {
+		return nil, 0, err
+	}
+
+	return cards, total, nil
+}
+
+// listStandaloneParallel 并行查询路径
+// 将 shop_id IN (...) 拆分为 per-shop 独立查询，每个查询走 Index Scan
+// 然后在应用层归并排序，避免 PG 对多值 IN + ORDER BY 选择全表扫描
+func (s *IotCardStore) listStandaloneParallel(ctx context.Context, opts *store.QueryOptions, filters map[string]any, shopIDs []uint) ([]*model.IotCard, int64, error) {
+	skipCtx := pkggorm.SkipDataPermission(ctx)
+
+	fetchLimit := (opts.Page-1)*opts.PageSize + opts.PageSize
+
+	// 深度分页（page >= 50）走两阶段延迟 Join：
+	// Phase 1 只取 ID（覆盖索引 Index Only Scan），Phase 2 用 ID 回表
+	// 避免 OFFSET 跳过大量宽行数据
+	if opts.Page >= 50 {
+		return s.listStandaloneParallelTwoPhase(ctx, opts, filters, shopIDs)
+	}
+
+	type shopResult struct {
+		cards []*model.IotCard
+		count int64
+		err   error
+	}
+
+	results := make([]shopResult, len(shopIDs))
+	var wg sync.WaitGroup
+
+	cachedTotal, hasCachedTotal := s.getCachedCount(ctx, "standalone", filters)
+
+	for i, shopID := range shopIDs {
+		wg.Add(1)
+		go func(idx int, sid uint) {
+			defer wg.Done()
+
+			q := s.db.WithContext(skipCtx).Model(&model.IotCard{}).
+				Where("is_standalone = true AND deleted_at IS NULL AND shop_id = ?", sid)
+			q = s.applyStandaloneFilters(skipCtx, q, filters)
+
+			var cards []*model.IotCard
+			if err := q.Select(standaloneListColumns).
+				Order("created_at DESC, id").
+				Limit(fetchLimit).
+				Find(&cards).Error; err != nil {
+				results[idx] = shopResult{err: err}
+				return
+			}
+
+			var count int64
+			if !hasCachedTotal {
+				countQ := s.db.WithContext(skipCtx).Model(&model.IotCard{}).
+					Where("is_standalone = true AND deleted_at IS NULL AND shop_id = ?", sid)
+				countQ = s.applyStandaloneFilters(skipCtx, countQ, filters)
+				if err := countQ.Count(&count).Error; err != nil {
+					results[idx] = shopResult{err: err}
+					return
+				}
+			}
+
+			results[idx] = shopResult{cards: cards, count: count}
+		}(i, shopID)
+	}
+
+	wg.Wait()
+
+	allCards := make([]*model.IotCard, 0, fetchLimit)
+	var totalCount int64
+	if hasCachedTotal {
+		totalCount = cachedTotal
+	}
+
+	for _, r := range results {
+		if r.err != nil {
+			return nil, 0, r.err
+		}
+		allCards = append(allCards, r.cards...)
+		if !hasCachedTotal {
+			totalCount += r.count
+		}
+	}
+
+	if !hasCachedTotal && totalCount > 0 {
+		s.cacheCount(ctx, "standalone", filters, totalCount)
+	}
+
+	sort.Slice(allCards, func(i, j int) bool {
+		if allCards[i].CreatedAt.Equal(allCards[j].CreatedAt) {
+			return allCards[i].ID < allCards[j].ID
+		}
+		return allCards[i].CreatedAt.After(allCards[j].CreatedAt)
+	})
+
+	offset := (opts.Page - 1) * opts.PageSize
+	end := offset + opts.PageSize
+	if offset >= len(allCards) {
+		return []*model.IotCard{}, totalCount, nil
+	}
+	if end > len(allCards) {
+		end = len(allCards)
+	}
+
+	logger.GetAppLogger().Debug("并行查询完成",
+		zap.Int("shop_count", len(shopIDs)),
+		zap.Int("total_fetched", len(allCards)),
+		zap.Int("offset", offset),
+		zap.Int("page_size", opts.PageSize),
+		zap.Int64("total_count", totalCount),
+	)
+
+	return allCards[offset:end], totalCount, nil
+}
+
+// cardIDWithTime 仅存储 ID 和排序键，用于两阶段查询的轻量归并排序
+type cardIDWithTime struct {
+	ID        uint
+	CreatedAt time.Time
+}
+
+// listStandaloneParallelTwoPhase 并行+两阶段延迟 Join 查询路径
+// 解决深度分页（page >= 50）时 OFFSET 跳过大量宽行数据的性能问题：
+// Phase 1: 每个 shop 并行查询 SELECT id, created_at（走覆盖索引 Index Only Scan，~20字节/元组）
+// 归并排序后取目标页的 20 个 ID
+// Phase 2: SELECT 完整列 WHERE id IN (20 IDs)（PK 精确回表）
+func (s *IotCardStore) listStandaloneParallelTwoPhase(ctx context.Context, opts *store.QueryOptions, filters map[string]any, shopIDs []uint) ([]*model.IotCard, int64, error) {
+	skipCtx := pkggorm.SkipDataPermission(ctx)
+
+	fetchLimit := (opts.Page-1)*opts.PageSize + opts.PageSize
+
+	type shopResult struct {
+		ids   []cardIDWithTime
+		count int64
+		err   error
+	}
+
+	results := make([]shopResult, len(shopIDs))
+	var wg sync.WaitGroup
+
+	cachedTotal, hasCachedTotal := s.getCachedCount(ctx, "standalone", filters)
+
+	// Phase 1: 并行获取每个 shop 的 (id, created_at)，走覆盖索引
+	for i, shopID := range shopIDs {
+		wg.Add(1)
+		go func(idx int, sid uint) {
+			defer wg.Done()
+
+			q := s.db.WithContext(skipCtx).Model(&model.IotCard{}).
+				Where("is_standalone = true AND deleted_at IS NULL AND shop_id = ?", sid)
+			q = s.applyStandaloneFilters(skipCtx, q, filters)
+
+			var ids []cardIDWithTime
+			if err := q.Select("id, created_at").
+				Order("created_at DESC, id").
+				Limit(fetchLimit).
+				Find(&ids).Error; err != nil {
+				results[idx] = shopResult{err: err}
+				return
+			}
+
+			var count int64
+			if !hasCachedTotal {
+				countQ := s.db.WithContext(skipCtx).Model(&model.IotCard{}).
+					Where("is_standalone = true AND deleted_at IS NULL AND shop_id = ?", sid)
+				countQ = s.applyStandaloneFilters(skipCtx, countQ, filters)
+				if err := countQ.Count(&count).Error; err != nil {
+					results[idx] = shopResult{err: err}
+					return
+				}
+			}
+
+			results[idx] = shopResult{ids: ids, count: count}
+		}(i, shopID)
+	}
+
+	wg.Wait()
+
+	allIDs := make([]cardIDWithTime, 0, fetchLimit)
+	var totalCount int64
+	if hasCachedTotal {
+		totalCount = cachedTotal
+	}
+
+	for _, r := range results {
+		if r.err != nil {
+			return nil, 0, r.err
+		}
+		allIDs = append(allIDs, r.ids...)
+		if !hasCachedTotal {
+			totalCount += r.count
+		}
+	}
+
+	if !hasCachedTotal && totalCount > 0 {
+		s.cacheCount(ctx, "standalone", filters, totalCount)
+	}
+
+	sort.Slice(allIDs, func(i, j int) bool {
+		if allIDs[i].CreatedAt.Equal(allIDs[j].CreatedAt) {
+			return allIDs[i].ID < allIDs[j].ID
+		}
+		return allIDs[i].CreatedAt.After(allIDs[j].CreatedAt)
+	})
+
+	offset := (opts.Page - 1) * opts.PageSize
+	end := offset + opts.PageSize
+	if offset >= len(allIDs) {
+		return []*model.IotCard{}, totalCount, nil
+	}
+	if end > len(allIDs) {
+		end = len(allIDs)
+	}
+
+	pageIDs := make([]uint, 0, opts.PageSize)
+	for _, item := range allIDs[offset:end] {
+		pageIDs = append(pageIDs, item.ID)
+	}
+
+	if len(pageIDs) == 0 {
+		return []*model.IotCard{}, totalCount, nil
+	}
+
+	// Phase 2: 用 ID 精确回表获取完整数据（PK Index Scan，仅 20 行）
+	var cards []*model.IotCard
+	if err := s.db.WithContext(skipCtx).Model(&model.IotCard{}).
+		Select(standaloneListColumns).
+		Where("id IN ?", pageIDs).
+		Find(&cards).Error; err != nil {
+		return nil, 0, err
+	}
+
+	// Phase 2 结果按原始排序顺序重排
+	idOrder := make(map[uint]int, len(pageIDs))
+	for i, id := range pageIDs {
+		idOrder[id] = i
+	}
+	sort.Slice(cards, func(i, j int) bool {
+		return idOrder[cards[i].ID] < idOrder[cards[j].ID]
+	})
+
+	logger.GetAppLogger().Debug("两阶段并行查询完成",
+		zap.Int("shop_count", len(shopIDs)),
+		zap.Int("phase1_total_ids", len(allIDs)),
+		zap.Int("phase2_fetched", len(cards)),
+		zap.Int("page", opts.Page),
+		zap.Int64("total_count", totalCount),
+	)
+
+	return cards, totalCount, nil
+}
+
+// applyStandaloneFilters 应用独立卡列表的通用过滤条件
+// 注意：不包含 is_standalone、shop_id、deleted_at 条件（由调用方控制）
+// 也不包含 subordinate_shop_ids（仅用于路由选择，不作为查询条件）
+func (s *IotCardStore) applyStandaloneFilters(ctx context.Context, query *gorm.DB, filters map[string]any) *gorm.DB {
+	skipCtx := pkggorm.SkipDataPermission(ctx)

 	if status, ok := filters["status"].(int); ok && status > 0 {
 		query = query.Where("status = ?", status)
@@ -198,6 +592,7 @@ func (s *IotCardStore) ListStandalone(ctx context.Context, opts *store.QueryOpti
 	if carrierID, ok := filters["carrier_id"].(uint); ok && carrierID > 0 {
 		query = query.Where("carrier_id = ?", carrierID)
 	}
+	// 并行路径下 shop_id 已由调用方设置，此处仅处理显式的 shop_id 过滤
 	if shopID, ok := filters["shop_id"].(uint); ok && shopID > 0 {
 		query = query.Where("shop_id = ?", shopID)
 	}
@@ -212,7 +607,7 @@ func (s *IotCardStore) ListStandalone(ctx context.Context, opts *store.QueryOpti
 	}
 	if packageID, ok := filters["package_id"].(uint); ok && packageID > 0 {
 		query = query.Where("id IN (?)",
-			s.db.Table("tb_package_usage").
+			s.db.WithContext(skipCtx).Table("tb_package_usage").
 				Select("iot_card_id").
 				Where("package_id = ? AND deleted_at IS NULL", packageID))
 	}
@@ -232,12 +627,12 @@ func (s *IotCardStore) ListStandalone(ctx context.Context, opts *store.QueryOpti
 	if isReplaced, ok := filters["is_replaced"].(bool); ok {
 		if isReplaced {
 			query = query.Where("id IN (?)",
-				s.db.Table("tb_card_replacement_record").
+				s.db.WithContext(skipCtx).Table("tb_card_replacement_record").
 					Select("old_iot_card_id").
 					Where("deleted_at IS NULL"))
 		} else {
 			query = query.Where("id NOT IN (?)",
-				s.db.Table("tb_card_replacement_record").
+				s.db.WithContext(skipCtx).Table("tb_card_replacement_record").
 					Select("old_iot_card_id").
 					Where("deleted_at IS NULL"))
 		}
@@ -246,30 +641,7 @@ func (s *IotCardStore) ListStandalone(ctx context.Context, opts *store.QueryOpti
 		query = query.Where("series_id = ?", seriesID)
 	}

-	if err := query.Count(&total).Error; err != nil {
-		return nil, 0, err
-	}
-
-	if opts == nil {
-		opts = &store.QueryOptions{
-			Page:     1,
-			PageSize: constants.DefaultPageSize,
-		}
-	}
-	offset := (opts.Page - 1) * opts.PageSize
-	query = query.Offset(offset).Limit(opts.PageSize)
-
-	if opts.OrderBy != "" {
-		query = query.Order(opts.OrderBy)
-	} else {
-		query = query.Order("created_at DESC")
-	}
-
-	if err := query.Find(&cards).Error; err != nil {
-		return nil, 0, err
-	}
-
-	return cards, total, nil
+	return query
 }

 func (s *IotCardStore) GetByICCIDs(ctx context.Context, iccids []string) ([]*model.IotCard, error) {
@@ -285,10 +657,7 @@ func (s *IotCardStore) GetByICCIDs(ctx context.Context, iccids []string) ([]*mod

 func (s *IotCardStore) GetStandaloneByICCIDRange(ctx context.Context, iccidStart, iccidEnd string, shopID *uint) ([]*model.IotCard, error) {
 	query := s.db.WithContext(ctx).Model(&model.IotCard{}).
-		Where("id NOT IN (?)",
-			s.db.Model(&model.DeviceSimBinding{}).
-				Select("iot_card_id").
-				Where("bind_status = ?", 1)).
+		Where("is_standalone = true").
 		Where("iccid >= ? AND iccid <= ?", iccidStart, iccidEnd)

 	if shopID == nil {
@@ -306,10 +675,7 @@ func (s *IotCardStore) GetStandaloneByICCIDRange(ctx context.Context, iccidStart

 func (s *IotCardStore) GetStandaloneByFilters(ctx context.Context, filters map[string]any, shopID *uint) ([]*model.IotCard, error) {
 	query := s.db.WithContext(ctx).Model(&model.IotCard{}).
-		Where("id NOT IN (?)",
-			s.db.Model(&model.DeviceSimBinding{}).
-				Select("iot_card_id").
-				Where("bind_status = ?", 1))
+		Where("is_standalone = true")

 	if shopID == nil {
 		query = query.Where("shop_id IS NULL")
@@ -365,9 +731,10 @@ func (s *IotCardStore) GetByIDsWithEnterpriseFilter(ctx context.Context, cardIDs
 	query := s.db.WithContext(ctx).Model(&model.IotCard{})

 	if enterpriseID != nil && *enterpriseID > 0 {
+		skipCtx := pkggorm.SkipDataPermission(ctx)
 		query = query.Where("id IN (?) AND id IN (?)",
 			cardIDs,
-			s.db.Table("tb_enterprise_card_authorization").
+			s.db.WithContext(skipCtx).Table("tb_enterprise_card_authorization").
 				Select("card_id").
 				Where("enterprise_id = ? AND revoked_at IS NULL AND deleted_at IS NULL", *enterpriseID))
 	} else {
@@ -475,3 +842,43 @@ func (s *IotCardStore) BatchDelete(ctx context.Context, cardIDs []uint) error {
 		Where("id IN ?", cardIDs).
 		Delete(&model.IotCard{}).Error
 }
+
+// ==================== 列表计数缓存 ====================
+
+func (s *IotCardStore) getCachedCount(ctx context.Context, table string, filters map[string]any) (int64, bool) {
+	if s.redis == nil {
+		return 0, false
+	}
+	key := constants.RedisListCountKey(table, middleware.GetUserIDFromContext(ctx), hashFilters(filters))
+	val, err := s.redis.Get(ctx, key).Int64()
+	if err != nil {
+		return 0, false
+	}
+	return val, true
+}
+
+func (s *IotCardStore) cacheCount(ctx context.Context, table string, filters map[string]any, total int64) {
+	if s.redis == nil {
+		return
+	}
+	key := constants.RedisListCountKey(table, middleware.GetUserIDFromContext(ctx), hashFilters(filters))
+	s.redis.Set(ctx, key, total, listCountCacheTTL)
+}
+
+func hashFilters(filters map[string]any) string {
+	if len(filters) == 0 {
+		return "0"
+	}
+	keys := make([]string, 0, len(filters))
+	for k := range filters {
+		keys = append(keys, k)
+	}
+	sort.Strings(keys)
+
+	h := fnv.New32a()
+	for _, k := range keys {
+		h.Write([]byte(k))
+		h.Write([]byte(fmt.Sprint(filters[k])))
+	}
+	return fmt.Sprintf("%08x", h.Sum32())
+}
--- a/migrations/000000_create_legacy_tables.down.sql
+++ b/migrations/000000_create_legacy_tables.down.sql
@@ -0,0 +1,5 @@
+-- 回滚：删除遗留表（逆序删除，先删关联表）
+DROP TABLE IF EXISTS tb_role_permission;
+DROP TABLE IF EXISTS tb_role;
+DROP TABLE IF EXISTS tb_permission;
+DROP TABLE IF EXISTS tb_account;
--- a/migrations/000000_create_legacy_tables.up.sql
+++ b/migrations/000000_create_legacy_tables.up.sql
@@ -0,0 +1,98 @@
+-- 创建遗留表（原由 GORM AutoMigrate 自动创建，迁移文件中缺失）
+-- 这 4 张表是项目最早期通过 AutoMigrate 创建的，后续迁移 000001-000010 依赖它们的存在
+-- 本迁移还原 AutoMigrate 创建时的原始结构（不含后续 ALTER 添加的字段）
+
+-- ============================================================
+-- 1. 账号表（tb_account）
+-- 注意：此处包含 parent_id 字段，该字段在 000002 迁移中被 DROP
+-- 注意：enterprise_id 由 000002 添加，is_primary 由 000010 添加，此处不包含
+-- ============================================================
+CREATE TABLE IF NOT EXISTS tb_account (
+    id BIGSERIAL PRIMARY KEY,
+    created_at TIMESTAMP WITH TIME ZONE,
+    updated_at TIMESTAMP WITH TIME ZONE,
+    deleted_at TIMESTAMP WITH TIME ZONE,
+    creator BIGINT NOT NULL,
+    updater BIGINT NOT NULL,
+    username VARCHAR(50) NOT NULL,
+    phone VARCHAR(20) NOT NULL,
+    password VARCHAR(255) NOT NULL,
+    user_type BIGINT NOT NULL,
+    shop_id BIGINT,
+    parent_id BIGINT,
+    status BIGINT NOT NULL DEFAULT 1
+);
+
+-- 索引（与 GORM AutoMigrate 生成的一致）
+CREATE UNIQUE INDEX IF NOT EXISTS idx_account_username ON tb_account(username) WHERE deleted_at IS NULL;
+CREATE UNIQUE INDEX IF NOT EXISTS idx_account_phone ON tb_account(phone) WHERE deleted_at IS NULL;
+CREATE INDEX IF NOT EXISTS idx_tb_account_user_type ON tb_account(user_type);
+CREATE INDEX IF NOT EXISTS idx_tb_account_shop_id ON tb_account(shop_id);
+CREATE INDEX IF NOT EXISTS idx_tb_account_deleted_at ON tb_account(deleted_at);
+
+-- ============================================================
+-- 2. 权限表（tb_permission）
+-- 注意：platform 由 000003 添加，available_for_role_types 由 000009 添加，此处不包含
+-- ============================================================
+CREATE TABLE IF NOT EXISTS tb_permission (
+    id BIGSERIAL PRIMARY KEY,
+    created_at TIMESTAMP WITH TIME ZONE,
+    updated_at TIMESTAMP WITH TIME ZONE,
+    deleted_at TIMESTAMP WITH TIME ZONE,
+    creator BIGINT NOT NULL,
+    updater BIGINT NOT NULL,
+    perm_name VARCHAR(50) NOT NULL,
+    perm_code VARCHAR(100) NOT NULL,
+    perm_type BIGINT NOT NULL,
+    url VARCHAR(255),
+    parent_id BIGINT,
+    sort BIGINT NOT NULL DEFAULT 0,
+    status BIGINT NOT NULL DEFAULT 1
+);
+
+-- 索引
+CREATE UNIQUE INDEX IF NOT EXISTS idx_permission_code ON tb_permission(perm_code) WHERE deleted_at IS NULL;
+CREATE INDEX IF NOT EXISTS idx_tb_permission_perm_type ON tb_permission(perm_type);
+CREATE INDEX IF NOT EXISTS idx_tb_permission_parent_id ON tb_permission(parent_id);
+CREATE INDEX IF NOT EXISTS idx_tb_permission_deleted_at ON tb_permission(deleted_at);
+
+-- ============================================================
+-- 3. 角色表（tb_role）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS tb_role (
+    id BIGSERIAL PRIMARY KEY,
+    created_at TIMESTAMP WITH TIME ZONE,
+    updated_at TIMESTAMP WITH TIME ZONE,
+    deleted_at TIMESTAMP WITH TIME ZONE,
+    creator BIGINT NOT NULL,
+    updater BIGINT NOT NULL,
+    role_name VARCHAR(50) NOT NULL,
+    role_desc VARCHAR(255),
+    role_type BIGINT NOT NULL,
+    status BIGINT NOT NULL DEFAULT 1
+);
+
+-- 索引
+CREATE INDEX IF NOT EXISTS idx_tb_role_role_type ON tb_role(role_type);
+CREATE INDEX IF NOT EXISTS idx_tb_role_deleted_at ON tb_role(deleted_at);
+
+-- ============================================================
+-- 4. 角色权限关联表（tb_role_permission）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS tb_role_permission (
+    id BIGSERIAL PRIMARY KEY,
+    created_at TIMESTAMP WITH TIME ZONE,
+    updated_at TIMESTAMP WITH TIME ZONE,
+    deleted_at TIMESTAMP WITH TIME ZONE,
+    creator BIGINT NOT NULL,
+    updater BIGINT NOT NULL,
+    role_id BIGINT NOT NULL,
+    perm_id BIGINT NOT NULL,
+    status BIGINT NOT NULL DEFAULT 1
+);
+
+-- 索引
+CREATE INDEX IF NOT EXISTS idx_tb_role_permission_role_id ON tb_role_permission(role_id);
+CREATE INDEX IF NOT EXISTS idx_tb_role_permission_perm_id ON tb_role_permission(perm_id);
+CREATE UNIQUE INDEX IF NOT EXISTS idx_role_permission_unique ON tb_role_permission(role_id, perm_id) WHERE deleted_at IS NULL;
+CREATE INDEX IF NOT EXISTS idx_tb_role_permission_deleted_at ON tb_role_permission(deleted_at);
--- a/migrations/000017_create_enterprise_card_authorization_table.up.sql
+++ b/migrations/000017_create_enterprise_card_authorization_table.up.sql
@@ -1,5 +1,7 @@
-- 创建企业卡授权表
-CREATE TABLE IF NOT EXISTS tb_enterprise_card_authorization (
+-- 重建企业卡授权表（替换 000010 中的旧版结构，字段从 iot_card_id 改为 card_id，移除 shop_id/status，新增 authorizer_type/revoked_by/revoked_at/remark）
+DROP TABLE IF EXISTS tb_enterprise_card_authorization;
+
+CREATE TABLE tb_enterprise_card_authorization (
    -- 基础字段
    id BIGSERIAL PRIMARY KEY,
    created_at TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
--- a/migrations/000056_add_is_standalone_to_iot_card.down.sql
+++ b/migrations/000056_add_is_standalone_to_iot_card.down.sql
@@ -0,0 +1,9 @@
+-- 回滚：删除 is_standalone 列和相关触发器
+
+DROP TRIGGER IF EXISTS trg_device_sim_binding_soft_delete ON tb_device_sim_binding;
+DROP TRIGGER IF EXISTS trg_device_sim_binding_standalone ON tb_device_sim_binding;
+DROP FUNCTION IF EXISTS fn_update_card_standalone_on_soft_delete();
+DROP FUNCTION IF EXISTS fn_update_card_standalone_on_bind();
+DROP INDEX IF EXISTS idx_iot_card_standalone_shop_status_created;
+DROP INDEX IF EXISTS idx_iot_card_standalone_shop_created;
+ALTER TABLE tb_iot_card DROP COLUMN IF EXISTS is_standalone;
--- a/migrations/000056_add_is_standalone_to_iot_card.up.sql
+++ b/migrations/000056_add_is_standalone_to_iot_card.up.sql
@@ -0,0 +1,79 @@
+-- 添加 is_standalone 物化列，消除 ListStandalone 查询中的 NOT EXISTS 子查询
+-- 30M 行表上 NOT EXISTS 子查询导致 9s+ 延迟，物化为布尔列后降至 <500ms
+
+-- 1. 添加列（默认 true，因为大多数卡未绑定设备）
+ALTER TABLE tb_iot_card ADD COLUMN IF NOT EXISTS is_standalone BOOLEAN NOT NULL DEFAULT true;
+
+COMMENT ON COLUMN tb_iot_card.is_standalone IS '是否为独立卡(未绑定设备) true=独立卡 false=已绑定设备，由触发器自动维护';
+
+-- 2. 回填数据：将已绑定设备的卡标记为 false
+UPDATE tb_iot_card SET is_standalone = false
+WHERE id IN (
+    SELECT DISTINCT iot_card_id FROM tb_device_sim_binding
+    WHERE bind_status = 1 AND deleted_at IS NULL
+);
+
+-- 3. 创建部分索引（仅包含独立卡 + 未删除的行）
+CREATE INDEX IF NOT EXISTS idx_iot_card_standalone_shop_created
+ON tb_iot_card (shop_id, created_at DESC)
+WHERE deleted_at IS NULL AND is_standalone = true;
+
+CREATE INDEX IF NOT EXISTS idx_iot_card_standalone_shop_status_created
+ON tb_iot_card (shop_id, status, created_at DESC)
+WHERE deleted_at IS NULL AND is_standalone = true;
+
+-- 4. 创建触发器函数：绑定时设置 is_standalone = false
+CREATE OR REPLACE FUNCTION fn_update_card_standalone_on_bind()
+RETURNS TRIGGER AS $$
+BEGIN
+    -- INSERT 或 bind_status 变为 1（绑定）
+    IF (TG_OP = 'INSERT' AND NEW.bind_status = 1) OR
+       (TG_OP = 'UPDATE' AND NEW.bind_status = 1 AND (OLD.bind_status IS DISTINCT FROM 1)) THEN
+        UPDATE tb_iot_card SET is_standalone = false WHERE id = NEW.iot_card_id;
+    END IF;
+
+    -- bind_status 从 1 变为其他值（解绑），检查是否还有其他活跃绑定
+    IF TG_OP = 'UPDATE' AND OLD.bind_status = 1 AND NEW.bind_status != 1 THEN
+        IF NOT EXISTS (
+            SELECT 1 FROM tb_device_sim_binding
+            WHERE iot_card_id = NEW.iot_card_id AND bind_status = 1
+              AND deleted_at IS NULL AND id != NEW.id
+        ) THEN
+            UPDATE tb_iot_card SET is_standalone = true WHERE id = NEW.iot_card_id;
+        END IF;
+    END IF;
+
+    RETURN NEW;
+END;
+$$ LANGUAGE plpgsql;
+
+-- 5. 绑定表的触发器
+CREATE TRIGGER trg_device_sim_binding_standalone
+AFTER INSERT OR UPDATE OF bind_status ON tb_device_sim_binding
+FOR EACH ROW
+EXECUTE FUNCTION fn_update_card_standalone_on_bind();
+
+-- 6. 软删除触发器：绑定记录被软删除时恢复独立状态
+CREATE OR REPLACE FUNCTION fn_update_card_standalone_on_soft_delete()
+RETURNS TRIGGER AS $$
+BEGIN
+    IF OLD.deleted_at IS NULL AND NEW.deleted_at IS NOT NULL AND OLD.bind_status = 1 THEN
+        IF NOT EXISTS (
+            SELECT 1 FROM tb_device_sim_binding
+            WHERE iot_card_id = NEW.iot_card_id AND bind_status = 1
+              AND deleted_at IS NULL AND id != NEW.id
+        ) THEN
+            UPDATE tb_iot_card SET is_standalone = true WHERE id = NEW.iot_card_id;
+        END IF;
+    END IF;
+    RETURN NEW;
+END;
+$$ LANGUAGE plpgsql;
+
+CREATE TRIGGER trg_device_sim_binding_soft_delete
+AFTER UPDATE OF deleted_at ON tb_device_sim_binding
+FOR EACH ROW
+EXECUTE FUNCTION fn_update_card_standalone_on_soft_delete();
+
+-- 7. 更新统计信息
+ANALYZE tb_iot_card;
--- a/migrations/000057_optimize_iot_card_indexes.down.sql
+++ b/migrations/000057_optimize_iot_card_indexes.down.sql
@@ -0,0 +1,8 @@
+-- 回滚：恢复全局索引，删除 carrier 复合索引
+DROP INDEX IF EXISTS idx_iot_card_standalone_shop_carrier_created;
+
+CREATE INDEX IF NOT EXISTS idx_iot_card_global_created_at
+ON tb_iot_card (created_at DESC)
+WHERE deleted_at IS NULL;
+
+ANALYZE tb_iot_card;
--- a/migrations/000057_optimize_iot_card_indexes.up.sql
+++ b/migrations/000057_optimize_iot_card_indexes.up.sql
@@ -0,0 +1,14 @@
+-- 优化 tb_iot_card 索引：修复 PG 优化器在 shop_id IN 场景下选错执行计划的问题
+-- 问题：idx_iot_card_global_created_at 导致优化器选择全表扫描+Filter，而不是更高效的 partial 索引
+-- 30M 行场景下 status+carrier_id 过滤查询从 4.5s 降至 <200ms
+
+-- 1. 删除全局 created_at 索引（诱导优化器走错计划的元凶）
+DROP INDEX IF EXISTS idx_iot_card_global_created_at;
+
+-- 2. 新建 carrier_id 复合 partial 索引（覆盖运营商+排序场景）
+CREATE INDEX IF NOT EXISTS idx_iot_card_standalone_shop_carrier_created
+ON tb_iot_card (shop_id, carrier_id, created_at DESC)
+WHERE deleted_at IS NULL AND is_standalone = true;
+
+-- 3. 更新统计信息
+ANALYZE tb_iot_card;
--- a/migrations/000058_add_covering_index_for_deep_pagination.down.sql
+++ b/migrations/000058_add_covering_index_for_deep_pagination.down.sql
@@ -0,0 +1,4 @@
+-- 回滚：删除深度分页覆盖索引
+DROP INDEX IF EXISTS idx_iot_card_standalone_shop_created_id;
+
+ANALYZE tb_iot_card;
--- a/migrations/000058_add_covering_index_for_deep_pagination.up.sql
+++ b/migrations/000058_add_covering_index_for_deep_pagination.up.sql
@@ -0,0 +1,15 @@
+-- 新增覆盖索引用于深度分页优化（两阶段延迟 Join）
+-- 问题：深度分页（page >= 50）需要 OFFSET 数千行，SELECT * 读取大量宽行数据（~2KB/行）
+-- 方案：Phase 1 仅扫描覆盖索引获取 ID，Phase 2 用 ID 批量回表取完整数据
+-- 预期：page 500 从 5.6s → <500ms
+--
+-- 生产环境建议手动执行带 CONCURRENTLY 的版本避免锁表：
+-- CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_iot_card_standalone_shop_created_id
+--   ON tb_iot_card (shop_id, created_at DESC, id)
+--   WHERE deleted_at IS NULL AND is_standalone = true;
+
+CREATE INDEX IF NOT EXISTS idx_iot_card_standalone_shop_created_id
+ON tb_iot_card (shop_id, created_at DESC, id)
+WHERE deleted_at IS NULL AND is_standalone = true;
+
+ANALYZE tb_iot_card;
--- a/opencode.json
+++ b/opencode.json
@@ -34,7 +34,7 @@
        "--access-mode=restricted"
      ],
      "environment": {
-        "DATABASE_URI": "postgresql://erp_pgsql:erp_2025@cxd.whcxd.cn:16159/junhong_cmp_test?sslmode=disable"
+        "DATABASE_URI": "postgresql://erp_pgsql:erp_2025@cxd.whcxd.cn:16289/junhong_cmp_test?sslmode=disable"
      }
    }
  }
--- a/pkg/constants/constants.go
+++ b/pkg/constants/constants.go
@@ -36,7 +36,7 @@ const (
 	DefaultConnMaxLifetime = 5 * time.Minute
 	DefaultPageSize        = 20
 	MaxPageSize            = 100
-	SlowQueryThreshold     = 100 * time.Millisecond
+	SlowQueryThreshold     = 500 * time.Millisecond
 )

 // 任务类型常量
--- a/pkg/constants/redis.go
+++ b/pkg/constants/redis.go
@@ -257,6 +257,17 @@ func RedisPackageActivationLockKey(carrierType string, carrierID uint) string {
 	return fmt.Sprintf("package:activation:lock:%s:%d", carrierType, carrierID)
 }

+// ========================================
+// 列表计数缓存 Redis Key
+// ========================================
+
+// RedisListCountKey 列表查询计数缓存键
+// 用途：缓存分页列表的 COUNT(*) 结果，避免每次翻页重复全表计数
+// 过期时间：30 秒
+func RedisListCountKey(table string, userID uint, filterHash string) string {
+	return fmt.Sprintf("list_count:%s:%d:%s", table, userID, filterHash)
+}
+
 // ========================================
 // 订单幂等性相关 Redis Key
 // ========================================
--- a/scripts/perf_query/bench.go
+++ b/scripts/perf_query/bench.go
@@ -0,0 +1,342 @@
+//go:build ignore
+
+// IoT 卡分页查询性能基准测试
+// 用法:
+//   go run ./scripts/perf_query/bench.go
+//   go run ./scripts/perf_query/bench.go -base-url http://localhost:3000 -n 20
+
+package main
+
+import (
+	"encoding/json"
+	"flag"
+	"fmt"
+	"io"
+	"log"
+	"net/http"
+	"os"
+	"sort"
+	"strings"
+	"time"
+)
+
+var (
+	baseURL  = flag.String("base-url", "", "API 服务地址 (默认读取 JUNHONG_SERVER_ADDRESS 环境变量)")
+	username = flag.String("username", "perf_test_agent", "登录用户名")
+	password = flag.String("password", "PerfTest@123456", "登录密码")
+	n        = flag.Int("n", 20, "每个场景执行次数")
+	warmup   = flag.Int("warmup", 3, "预热请求次数")
+)
+
+type apiResponse struct {
+	Code int             `json:"code"`
+	Msg  string          `json:"msg"`
+	Data json.RawMessage `json:"data"`
+}
+
+type loginData struct {
+	AccessToken string `json:"access_token"`
+}
+
+type listData struct {
+	Total    int64 `json:"total"`
+	Page     int   `json:"page"`
+	PageSize int   `json:"page_size"`
+}
+
+type scenario struct {
+	Name   string
+	Path   string
+	Params string
+}
+
+func main() {
+	flag.Parse()
+
+	if *baseURL == "" {
+		addr := os.Getenv("JUNHONG_SERVER_ADDRESS")
+		if addr == "" {
+			addr = ":3000"
+		}
+		if strings.HasPrefix(addr, ":") {
+			addr = "http://localhost" + addr
+		}
+		*baseURL = addr
+	}
+
+	fmt.Println("╔══════════════════════════════════════════════════════════════╗")
+	fmt.Println("║          IoT 卡分页查询性能基准测试                         ║")
+	fmt.Println("╚══════════════════════════════════════════════════════════════╝")
+	fmt.Printf("  服务地址: %s\n", *baseURL)
+	fmt.Printf("  用户: %s\n", *username)
+	fmt.Printf("  每场景次数: %d (预热 %d 次)\n", *n, *warmup)
+	fmt.Println()
+
+	token := login()
+	fmt.Printf("✅ 登录成功 (token: %s...)\n\n", token[:20])
+
+	scenarios := []scenario{
+		{
+			Name:   "A: 无过滤分页 (第1页)",
+			Path:   "/api/admin/iot-cards/standalone",
+			Params: "page=1&page_size=20",
+		},
+		{
+			Name:   "B: status+carrier 过滤",
+			Path:   "/api/admin/iot-cards/standalone",
+			Params: "page=1&page_size=20&status=3&carrier_id=2",
+		},
+		{
+			Name:   "C: 深分页 (第500页)",
+			Path:   "/api/admin/iot-cards/standalone",
+			Params: "page=500&page_size=20",
+		},
+		{
+			Name:   "D: ICCID 模糊搜索",
+			Path:   "/api/admin/iot-cards/standalone",
+			Params: "page=1&page_size=20&iccid=12345",
+		},
+		{
+			Name:   "E: 批次号精确过滤",
+			Path:   "/api/admin/iot-cards/standalone",
+			Params: "page=1&page_size=20&batch_no=PERF-TEST-30M",
+		},
+		{
+			Name:   "F: 运营商+状态+分页",
+			Path:   "/api/admin/iot-cards/standalone",
+			Params: "page=10&page_size=50&status=1&carrier_id=1",
+		},
+	}
+
+	results := make([]benchResult, 0, len(scenarios))
+
+	for _, sc := range scenarios {
+		r := runScenario(token, sc)
+		results = append(results, r)
+	}
+
+	printSummary(results)
+}
+
+type benchResult struct {
+	Name     string
+	Total    int64
+	Timings  []time.Duration
+	Min      time.Duration
+	Max      time.Duration
+	Avg      time.Duration
+	P50      time.Duration
+	P95      time.Duration
+	P99      time.Duration
+	Pass     bool
+	ErrorMsg string
+}
+
+func runScenario(token string, sc scenario) benchResult {
+	fmt.Printf("🔄 %s\n", sc.Name)
+	fmt.Printf("   URL: %s?%s\n", sc.Path, sc.Params)
+
+	url := fmt.Sprintf("%s%s?%s", *baseURL, sc.Path, sc.Params)
+
+	result := benchResult{Name: sc.Name}
+
+	for i := 0; i < *warmup; i++ {
+		dur, _, err := doRequest(url, token)
+		if err != nil {
+			result.ErrorMsg = fmt.Sprintf("预热失败: %v", err)
+			fmt.Printf("   ❌ %s\n\n", result.ErrorMsg)
+			return result
+		}
+		fmt.Printf("   预热 %d/%d: %v\n", i+1, *warmup, dur.Round(time.Millisecond))
+	}
+
+	timings := make([]time.Duration, 0, *n)
+	var totalCount int64
+
+	for i := 0; i < *n; i++ {
+		dur, data, err := doRequest(url, token)
+		if err != nil {
+			result.ErrorMsg = fmt.Sprintf("第 %d 次请求失败: %v", i+1, err)
+			fmt.Printf("   ❌ %s\n\n", result.ErrorMsg)
+			return result
+		}
+		timings = append(timings, dur)
+		if i == 0 {
+			totalCount = data.Total
+		}
+		if (i+1)%5 == 0 || i == 0 {
+			fmt.Printf("   [%d/%d] %v\n", i+1, *n, dur.Round(time.Millisecond))
+		}
+	}
+
+	result.Total = totalCount
+	result.Timings = timings
+	result.Pass = true
+	calcStats(&result)
+
+	passStr := "✅"
+	if result.P95 > 1500*time.Millisecond {
+		passStr = "❌"
+		result.Pass = false
+	} else if result.P95 > 1000*time.Millisecond {
+		passStr = "⚠️"
+	}
+
+	fmt.Printf("   匹配行数: %d\n", result.Total)
+	fmt.Printf("   %s P50=%v  P95=%v  P99=%v  (min=%v max=%v)\n\n",
+		passStr, result.P50, result.P95, result.P99, result.Min, result.Max)
+
+	return result
+}
+
+func doRequest(url, token string) (time.Duration, *listData, error) {
+	req, _ := http.NewRequest("GET", url, nil)
+	req.Header.Set("Authorization", "Bearer "+token)
+
+	start := time.Now()
+	resp, err := http.DefaultClient.Do(req)
+	dur := time.Since(start)
+
+	if err != nil {
+		return dur, nil, fmt.Errorf("HTTP 请求失败: %w", err)
+	}
+	defer resp.Body.Close()
+
+	body, _ := io.ReadAll(resp.Body)
+
+	if resp.StatusCode != 200 {
+		return dur, nil, fmt.Errorf("HTTP %d: %s", resp.StatusCode, string(body[:min(len(body), 200)]))
+	}
+
+	var apiResp apiResponse
+	if err := json.Unmarshal(body, &apiResp); err != nil {
+		return dur, nil, fmt.Errorf("JSON 解析失败: %w", err)
+	}
+
+	if apiResp.Code != 0 {
+		return dur, nil, fmt.Errorf("API 错误 code=%d: %s", apiResp.Code, apiResp.Msg)
+	}
+
+	var data listData
+	json.Unmarshal(apiResp.Data, &data)
+
+	return dur, &data, nil
+}
+
+func login() string {
+	url := *baseURL + "/api/auth/login"
+	payload := fmt.Sprintf(`{"username":"%s","password":"%s","device":"web"}`, *username, *password)
+
+	resp, err := http.Post(url, "application/json", strings.NewReader(payload))
+	if err != nil {
+		log.Fatalf("❌ 登录请求失败: %v", err)
+	}
+	defer resp.Body.Close()
+
+	body, _ := io.ReadAll(resp.Body)
+
+	if resp.StatusCode != 200 {
+		log.Fatalf("❌ 登录 HTTP %d: %s", resp.StatusCode, string(body[:min(len(body), 500)]))
+	}
+
+	var apiResp apiResponse
+	if err := json.Unmarshal(body, &apiResp); err != nil {
+		log.Fatalf("❌ 登录响应解析失败: %v", err)
+	}
+
+	if apiResp.Code != 0 {
+		log.Fatalf("❌ 登录失败 code=%d: %s", apiResp.Code, apiResp.Msg)
+	}
+
+	var data loginData
+	if err := json.Unmarshal(apiResp.Data, &data); err != nil {
+		log.Fatalf("❌ 解析 token 失败: %v", err)
+	}
+
+	if data.AccessToken == "" {
+		log.Fatalf("❌ 登录成功但 token 为空")
+	}
+
+	return data.AccessToken
+}
+
+func calcStats(r *benchResult) {
+	sorted := make([]time.Duration, len(r.Timings))
+	copy(sorted, r.Timings)
+	sort.Slice(sorted, func(i, j int) bool { return sorted[i] < sorted[j] })
+
+	var sum time.Duration
+	for _, d := range sorted {
+		sum += d
+	}
+
+	n := len(sorted)
+	r.Min = sorted[0]
+	r.Max = sorted[n-1]
+	r.Avg = sum / time.Duration(n)
+	r.P50 = percentile(sorted, 50)
+	r.P95 = percentile(sorted, 95)
+	r.P99 = percentile(sorted, 99)
+}
+
+func percentile(sorted []time.Duration, pct int) time.Duration {
+	if len(sorted) == 0 {
+		return 0
+	}
+	idx := (pct * len(sorted)) / 100
+	if idx >= len(sorted) {
+		idx = len(sorted) - 1
+	}
+	return sorted[idx]
+}
+
+func printSummary(results []benchResult) {
+	fmt.Println("╔══════════════════════════════════════════════════════════════════════════════╗")
+	fmt.Println("║                              测试结果汇总                                   ║")
+	fmt.Println("╠══════════════════════════════════════════════════════════════════════════════╣")
+	fmt.Printf("║ %-32s │ %8s │ %8s │ %8s │ %8s ║\n", "场景", "匹配行数", "P50", "P95", "结果")
+	fmt.Println("╠══════════════════════════════════════════════════════════════════════════════╣")
+
+	allPass := true
+	for _, r := range results {
+		status := "✅ PASS"
+		if r.ErrorMsg != "" {
+			status = "❌ ERROR"
+			allPass = false
+		} else if !r.Pass {
+			status = "❌ FAIL"
+			allPass = false
+		} else if r.P95 > 1000*time.Millisecond {
+			status = "⚠️ SLOW"
+		}
+
+		p50Str := "-"
+		p95Str := "-"
+		totalStr := "-"
+		if r.ErrorMsg == "" {
+			p50Str = r.P50.Round(time.Millisecond).String()
+			p95Str = r.P95.Round(time.Millisecond).String()
+			totalStr = fmt.Sprintf("%d", r.Total)
+		}
+
+		fmt.Printf("║ %-32s │ %8s │ %8s │ %8s │ %8s ║\n",
+			r.Name, totalStr, p50Str, p95Str, status)
+	}
+
+	fmt.Println("╚══════════════════════════════════════════════════════════════════════════════╝")
+
+	if allPass {
+		fmt.Println("\n🎉 所有场景 P95 < 1.5s，满足性能要求!")
+	} else {
+		fmt.Println("\n⚠️  部分场景未达标，建议:")
+		fmt.Println("   1. 运行 go run ./scripts/perf_query/seed.go -action add-index 创建优化索引")
+		fmt.Println("   2. 重新运行本测试对比效果")
+	}
+}
+
+func min(a, b int) int {
+	if a < b {
+		return a
+	}
+	return b
+}
--- a/scripts/perf_query/deploy-perf-pg.sh
+++ b/scripts/perf_query/deploy-perf-pg.sh
@@ -0,0 +1,85 @@
+#!/bin/bash
+set -e
+
+DATA_DIR="/mnt/data1/database/pgsql/pgdata_perf"
+CONTAINER="postgres-perf-test"
+
+echo "=========================================="
+echo " 性能测试 PostgreSQL 容器部署"
+echo "=========================================="
+
+if [ -d "$DATA_DIR" ] && [ "$(ls -A $DATA_DIR 2>/dev/null)" ]; then
+    echo "⚠️  数据目录已存在"
+    read -p "清空重建？(y/N): " confirm
+    if [ "$confirm" = "y" ] || [ "$confirm" = "Y" ]; then
+        docker stop $CONTAINER 2>/dev/null || true
+        docker rm $CONTAINER 2>/dev/null || true
+        sudo rm -rf "$DATA_DIR"
+    fi
+fi
+
+sudo mkdir -p "$DATA_DIR"
+sudo chown -R 999:999 "$DATA_DIR"
+
+# 第一步：用默认配置启动，让 PG 完成初始化
+echo "启动容器（默认配置初始化）..."
+docker compose -f docker-compose-perf.yml up -d
+
+echo "等待初始化完成..."
+for i in $(seq 1 30); do
+    if docker exec $CONTAINER pg_isready -U erp_pgsql -d junhong_cmp_test > /dev/null 2>&1; then
+        echo "✅ 初始化完成"
+        break
+    fi
+    [ $i -eq 30 ] && echo "❌ 超时" && docker logs $CONTAINER && exit 1
+    sleep 2
+done
+
+# 第二步：把优化后的配置覆盖到数据目录
+echo "注入优化配置..."
+docker cp postgresql-perf.conf $CONTAINER:/var/lib/postgresql/data/postgresql.conf
+
+# 第三步：重启让新配置生效
+echo "重启容器..."
+docker restart $CONTAINER
+
+echo "等待重启完成..."
+for i in $(seq 1 30); do
+    if docker exec $CONTAINER pg_isready -U erp_pgsql -d junhong_cmp_test > /dev/null 2>&1; then
+        echo "✅ 重启完成"
+        break
+    fi
+    [ $i -eq 30 ] && echo "❌ 超时" && docker logs --tail 20 $CONTAINER && exit 1
+    sleep 2
+done
+
+# 第四步：启用扩展
+echo "启用扩展..."
+docker exec $CONTAINER psql -U erp_pgsql -d junhong_cmp_test -c "CREATE EXTENSION IF NOT EXISTS pg_stat_statements;" 2>/dev/null || echo "pg_stat_statements 需要 shared_preload_libraries，跳过"
+docker exec $CONTAINER psql -U erp_pgsql -d junhong_cmp_test -c "CREATE EXTENSION IF NOT EXISTS pg_trgm;"
+
+# 验证
+echo ""
+echo "=========================================="
+echo " 配置验证"
+echo "=========================================="
+docker exec $CONTAINER psql -U erp_pgsql -d junhong_cmp_test -c "
+SELECT name, setting, unit
+FROM pg_settings
+WHERE name IN (
+    'shared_buffers', 'effective_cache_size', 'work_mem',
+    'random_page_cost', 'effective_io_concurrency',
+    'max_parallel_workers_per_gather', 'maintenance_work_mem',
+    'max_connections', 'autovacuum_vacuum_scale_factor'
+)
+ORDER BY name;
+"
+
+echo ""
+echo "=========================================="
+echo " 部署完成"
+echo "=========================================="
+echo "  Port:     16289"
+echo "  Database: junhong_cmp_test"
+echo "  User:     erp_pgsql / erp_2025"
+echo ""
--- a/scripts/perf_query/docker-compose-perf.yml
+++ b/scripts/perf_query/docker-compose-perf.yml
@@ -0,0 +1,21 @@
+version: '3.8'
+
+services:
+  postgres-perf:
+    image: postgres:latest
+    container_name: postgres-perf-test
+    restart: unless-stopped
+    ports:
+      - "16289:5432"
+    environment:
+      POSTGRES_USER: erp_pgsql
+      POSTGRES_PASSWORD: erp_2025
+      POSTGRES_DB: junhong_cmp_test
+      TZ: Asia/Shanghai
+    volumes:
+      - /mnt/data1/database/pgsql/pgdata_perf:/var/lib/postgresql/data
+    shm_size: '18g'
+    deploy:
+      resources:
+        limits:
+          memory: 32G
--- a/scripts/perf_query/postgresql-perf.conf
+++ b/scripts/perf_query/postgresql-perf.conf
@@ -0,0 +1,840 @@
+# -----------------------------
+# PostgreSQL configuration file
+# -----------------------------
+#
+# This file consists of lines of the form:
+#
+#   name = value
+#
+# (The "=" is optional.)  Whitespace may be used.  Comments are introduced with
+# "#" anywhere on a line.  The complete list of parameter names and allowed
+# values can be found in the PostgreSQL documentation.
+#
+# The commented-out settings shown in this file represent the default values.
+# Re-commenting a setting is NOT sufficient to revert it to the default value;
+# you need to reload the server.
+#
+# This file is read on server startup and when the server receives a SIGHUP
+# signal.  If you edit the file on a running system, you have to SIGHUP the
+# server for the changes to take effect, run "pg_ctl reload", or execute
+# "SELECT pg_reload_conf()".  Some parameters, which are marked below,
+# require a server shutdown and restart to take effect.
+#
+# Any parameter can also be given as a command-line option to the server, e.g.,
+# "postgres -c log_connections=on".  Some parameters can be changed at run time
+# with the "SET" SQL command.
+#
+# Memory units:  B  = bytes            Time units:  us  = microseconds
+#                kB = kilobytes                     ms  = milliseconds
+#                MB = megabytes                     s   = seconds
+#                GB = gigabytes                     min = minutes
+#                TB = terabytes                     h   = hours
+#                                                   d   = days
+
+
+#------------------------------------------------------------------------------
+# FILE LOCATIONS
+#------------------------------------------------------------------------------
+
+# The default values of these variables are driven from the -D command-line
+# option or PGDATA environment variable, represented here as ConfigDir.
+
+#data_directory = 'ConfigDir'		# use data in another directory
+					# (change requires restart)
+#hba_file = 'ConfigDir/pg_hba.conf'	# host-based authentication file
+					# (change requires restart)
+#ident_file = 'ConfigDir/pg_ident.conf'	# ident configuration file
+					# (change requires restart)
+
+# If external_pid_file is not explicitly set, no extra PID file is written.
+#external_pid_file = ''			# write an extra PID file
+					# (change requires restart)
+
+
+#------------------------------------------------------------------------------
+# CONNECTIONS AND AUTHENTICATION
+#------------------------------------------------------------------------------
+
+# - Connection Settings -
+
+listen_addresses = '*'
+					# comma-separated list of addresses;
+					# defaults to 'localhost'; use '*' for all
+					# (change requires restart)
+#port = 5432				# (change requires restart)
+max_connections = 200			# (change requires restart)
+#reserved_connections = 0		# (change requires restart)
+#superuser_reserved_connections = 3	# (change requires restart)
+#unix_socket_directories = '/var/run/postgresql' # comma-separated list of directories
+					# (change requires restart)
+#unix_socket_group = ''			# (change requires restart)
+#unix_socket_permissions = 0777		# begin with 0 to use octal notation
+					# (change requires restart)
+#bonjour = off				# advertise server via Bonjour
+					# (change requires restart)
+#bonjour_name = ''			# defaults to the computer name
+					# (change requires restart)
+
+# - TCP settings -
+# see "man tcp" for details
+
+#tcp_keepalives_idle = 0		# TCP_KEEPIDLE, in seconds;
+					# 0 selects the system default
+#tcp_keepalives_interval = 0		# TCP_KEEPINTVL, in seconds;
+					# 0 selects the system default
+#tcp_keepalives_count = 0		# TCP_KEEPCNT;
+					# 0 selects the system default
+#tcp_user_timeout = 0			# TCP_USER_TIMEOUT, in milliseconds;
+					# 0 selects the system default
+
+#client_connection_check_interval = 0	# time between checks for client
+					# disconnection while running queries;
+					# 0 for never
+
+# - Authentication -
+
+#authentication_timeout = 1min		# 1s-600s
+#password_encryption = scram-sha-256	# scram-sha-256 or md5
+#scram_iterations = 4096
+
+# GSSAPI using Kerberos
+#krb_server_keyfile = 'FILE:${sysconfdir}/krb5.keytab'
+#krb_caseins_users = off
+#gss_accept_delegation = off
+
+# - SSL -
+
+#ssl = off
+#ssl_ca_file = ''
+#ssl_cert_file = 'server.crt'
+#ssl_crl_file = ''
+#ssl_crl_dir = ''
+#ssl_key_file = 'server.key'
+#ssl_ciphers = 'HIGH:MEDIUM:+3DES:!aNULL'	# allowed SSL ciphers
+#ssl_prefer_server_ciphers = on
+#ssl_ecdh_curve = 'prime256v1'
+#ssl_min_protocol_version = 'TLSv1.2'
+#ssl_max_protocol_version = ''
+#ssl_dh_params_file = ''
+#ssl_passphrase_command = ''
+#ssl_passphrase_command_supports_reload = off
+
+
+#------------------------------------------------------------------------------
+# RESOURCE USAGE (except WAL)
+#------------------------------------------------------------------------------
+
+# - Memory -
+
+shared_buffers = 16GB			# 服务器 125GB RAM，取 ~13%（保守值，与其他服务共存）
+					# (change requires restart)
+huge_pages = try			# 16GB shared_buffers 建议开启大页，减少 TLB miss
+					# (change requires restart)
+#huge_page_size = 0			# zero for system default
+					# (change requires restart)
+#temp_buffers = 8MB			# min 800kB
+#max_prepared_transactions = 0		# zero disables the feature
+					# (change requires restart)
+# Caution: it is not advisable to set max_prepared_transactions nonzero unless
+# you actively intend to use prepared transactions.
+work_mem = 128MB			# 排序/哈希在内存完成，避免溢出磁盘（默认 4MB 太小）
+#hash_mem_multiplier = 2.0		# 1-1000.0 multiplier on hash table work_mem
+maintenance_work_mem = 2GB		# VACUUM/CREATE INDEX 加速（默认 64MB 太小）
+#autovacuum_work_mem = -1		# min 64kB, or -1 to use maintenance_work_mem
+#logical_decoding_work_mem = 64MB	# min 64kB
+#max_stack_depth = 2MB			# min 100kB
+#shared_memory_type = mmap		# the default is the first option
+					# supported by the operating system:
+					#   mmap
+					#   sysv
+					#   windows
+					# (change requires restart)
+dynamic_shared_memory_type = posix	# the default is usually the first option
+					# supported by the operating system:
+					#   posix
+					#   sysv
+					#   windows
+					#   mmap
+					# (change requires restart)
+#min_dynamic_shared_memory = 0MB	# (change requires restart)
+#vacuum_buffer_usage_limit = 2MB	# size of vacuum and analyze buffer access strategy ring;
+					# 0 to disable vacuum buffer access strategy;
+					# range 128kB to 16GB
+
+# SLRU buffers (change requires restart)
+#commit_timestamp_buffers = 0		# memory for pg_commit_ts (0 = auto)
+#multixact_offset_buffers = 16		# memory for pg_multixact/offsets
+#multixact_member_buffers = 32		# memory for pg_multixact/members
+#notify_buffers = 16			# memory for pg_notify
+#serializable_buffers = 32		# memory for pg_serial
+#subtransaction_buffers = 0		# memory for pg_subtrans (0 = auto)
+#transaction_buffers = 0		# memory for pg_xact (0 = auto)
+
+# - Disk -
+
+#temp_file_limit = -1			# limits per-process temp file space
+					# in kilobytes, or -1 for no limit
+
+#max_notify_queue_pages = 1048576	# limits the number of SLRU pages allocated
+					# for NOTIFY / LISTEN queue
+
+# - Kernel Resources -
+
+#max_files_per_process = 1000		# min 64
+					# (change requires restart)
+
+# - Cost-Based Vacuum Delay -
+
+#vacuum_cost_delay = 0			# 0-100 milliseconds (0 disables)
+#vacuum_cost_page_hit = 1		# 0-10000 credits
+#vacuum_cost_page_miss = 2		# 0-10000 credits
+#vacuum_cost_page_dirty = 20		# 0-10000 credits
+#vacuum_cost_limit = 200		# 1-10000 credits
+
+# - Background Writer -
+
+#bgwriter_delay = 200ms			# 10-10000ms between rounds
+#bgwriter_lru_maxpages = 100		# max buffers written/round, 0 disables
+#bgwriter_lru_multiplier = 2.0		# 0-10.0 multiplier on buffers scanned/round
+#bgwriter_flush_after = 512kB		# measured in pages, 0 disables
+
+# - Asynchronous Behavior -
+
+#backend_flush_after = 0		# measured in pages, 0 disables
+effective_io_concurrency = 4		# HDD + RAID 控制器（MR9362-8i），支持少量并发预读
+maintenance_io_concurrency = 4		# VACUUM 预读并发，与 effective_io_concurrency 一致
+#io_combine_limit = 128kB		# usually 1-32 blocks (depends on OS)
+#max_worker_processes = 8		# (change requires restart)
+max_parallel_workers_per_gather = 4	# 大表查询多用几个核并行扫描
+max_parallel_maintenance_workers = 4	# VACUUM/CREATE INDEX 并行加速
+max_parallel_workers = 8		# number of max_worker_processes that
+					# can be used in parallel operations
+#parallel_leader_participation = on
+
+
+#------------------------------------------------------------------------------
+# WRITE-AHEAD LOG
+#------------------------------------------------------------------------------
+
+# - Settings -
+
+#wal_level = replica			# minimal, replica, or logical
+					# (change requires restart)
+#fsync = on				# flush data to disk for crash safety
+					# (turning this off can cause
+					# unrecoverable data corruption)
+#synchronous_commit = on		# synchronization level;
+					# off, local, remote_write, remote_apply, or on
+#wal_sync_method = fsync		# the default is the first option
+					# supported by the operating system:
+					#   open_datasync
+					#   fdatasync (default on Linux and FreeBSD)
+					#   fsync
+					#   fsync_writethrough
+					#   open_sync
+#full_page_writes = on			# recover from partial page writes
+#wal_log_hints = off			# also do full page writes of non-critical updates
+					# (change requires restart)
+#wal_compression = off			# enables compression of full-page writes;
+					# off, pglz, lz4, zstd, or on
+#wal_init_zero = on			# zero-fill new WAL files
+#wal_recycle = on			# recycle WAL files
+#wal_buffers = -1			# min 32kB, -1 sets based on shared_buffers
+					# (change requires restart)
+#wal_writer_delay = 200ms		# 1-10000 milliseconds
+#wal_writer_flush_after = 1MB		# measured in pages, 0 disables
+#wal_skip_threshold = 2MB
+
+#commit_delay = 0			# range 0-100000, in microseconds
+#commit_siblings = 5			# range 1-1000
+
+# - Checkpoints -
+
+checkpoint_timeout = 15min		# 拉长 checkpoint 间隔，减少写 I/O
+checkpoint_completion_target = 0.9	# checkpoint target duration, 0.0 - 1.0
+#checkpoint_flush_after = 256kB		# measured in pages, 0 disables
+#checkpoint_warning = 30s		# 0 disables
+max_wal_size = 4GB			# 减少 checkpoint 频率，降低 I/O 尖峰
+min_wal_size = 1GB
+
+# - Prefetching during recovery -
+
+#recovery_prefetch = try	# prefetch pages referenced in the WAL?
+#wal_decode_buffer_size = 512kB	# lookahead window used for prefetching
+				# (change requires restart)
+
+# - Archiving -
+
+#archive_mode = off		# enables archiving; off, on, or always
+				# (change requires restart)
+#archive_library = ''		# library to use to archive a WAL file
+				# (empty string indicates archive_command should
+				# be used)
+#archive_command = ''		# command to use to archive a WAL file
+				# placeholders: %p = path of file to archive
+				#               %f = file name only
+				# e.g. 'test ! -f /mnt/server/archivedir/%f && cp %p /mnt/server/archivedir/%f'
+#archive_timeout = 0		# force a WAL file switch after this
+				# number of seconds; 0 disables
+
+# - Archive Recovery -
+
+# These are only used in recovery mode.
+
+#restore_command = ''		# command to use to restore an archived WAL file
+				# placeholders: %p = path of file to restore
+				#               %f = file name only
+				# e.g. 'cp /mnt/server/archivedir/%f %p'
+#archive_cleanup_command = ''	# command to execute at every restartpoint
+#recovery_end_command = ''	# command to execute at completion of recovery
+
+# - Recovery Target -
+
+# Set these only when performing a targeted recovery.
+
+#recovery_target = ''		# 'immediate' to end recovery as soon as a
+				# consistent state is reached
+				# (change requires restart)
+#recovery_target_name = ''	# the named restore point to which recovery will proceed
+				# (change requires restart)
+#recovery_target_time = ''	# the time stamp up to which recovery will proceed
+				# (change requires restart)
+#recovery_target_xid = ''	# the transaction ID up to which recovery will proceed
+				# (change requires restart)
+#recovery_target_lsn = ''	# the WAL LSN up to which recovery will proceed
+				# (change requires restart)
+#recovery_target_inclusive = on	# Specifies whether to stop:
+				# just after the specified recovery target (on)
+				# just before the recovery target (off)
+				# (change requires restart)
+#recovery_target_timeline = 'latest'	# 'current', 'latest', or timeline ID
+					# (change requires restart)
+#recovery_target_action = 'pause'	# 'pause', 'promote', 'shutdown'
+					# (change requires restart)
+
+# - WAL Summarization -
+
+#summarize_wal = off		# run WAL summarizer process?
+#wal_summary_keep_time = '10d'	# when to remove old summary files, 0 = never
+
+
+#------------------------------------------------------------------------------
+# REPLICATION
+#------------------------------------------------------------------------------
+
+# - Sending Servers -
+
+# Set these on the primary and on any standby that will send replication data.
+
+#max_wal_senders = 10		# max number of walsender processes
+				# (change requires restart)
+#max_replication_slots = 10	# max number of replication slots
+				# (change requires restart)
+#wal_keep_size = 0		# in megabytes; 0 disables
+#max_slot_wal_keep_size = -1	# in megabytes; -1 disables
+#wal_sender_timeout = 60s	# in milliseconds; 0 disables
+#track_commit_timestamp = off	# collect timestamp of transaction commit
+				# (change requires restart)
+
+# - Primary Server -
+
+# These settings are ignored on a standby server.
+
+#synchronous_standby_names = ''	# standby servers that provide sync rep
+				# method to choose sync standbys, number of sync standbys,
+				# and comma-separated list of application_name
+				# from standby(s); '*' = all
+#synchronized_standby_slots = ''	# streaming replication standby server slot
+				# names that logical walsender processes will wait for
+
+# - Standby Servers -
+
+# These settings are ignored on a primary server.
+
+#primary_conninfo = ''			# connection string to sending server
+#primary_slot_name = ''			# replication slot on sending server
+#hot_standby = on			# "off" disallows queries during recovery
+					# (change requires restart)
+#max_standby_archive_delay = 30s	# max delay before canceling queries
+					# when reading WAL from archive;
+					# -1 allows indefinite delay
+#max_standby_streaming_delay = 30s	# max delay before canceling queries
+					# when reading streaming WAL;
+					# -1 allows indefinite delay
+#wal_receiver_create_temp_slot = off	# create temp slot if primary_slot_name
+					# is not set
+#wal_receiver_status_interval = 10s	# send replies at least this often
+					# 0 disables
+#hot_standby_feedback = off		# send info from standby to prevent
+					# query conflicts
+#wal_receiver_timeout = 60s		# time that receiver waits for
+					# communication from primary
+					# in milliseconds; 0 disables
+#wal_retrieve_retry_interval = 5s	# time to wait before retrying to
+					# retrieve WAL after a failed attempt
+#recovery_min_apply_delay = 0		# minimum delay for applying changes during recovery
+#sync_replication_slots = off		# enables slot synchronization on the physical standby from the primary
+
+# - Subscribers -
+
+# These settings are ignored on a publisher.
+
+#max_logical_replication_workers = 4	# taken from max_worker_processes
+					# (change requires restart)
+#max_sync_workers_per_subscription = 2	# taken from max_logical_replication_workers
+#max_parallel_apply_workers_per_subscription = 2	# taken from max_logical_replication_workers
+
+
+#------------------------------------------------------------------------------
+# QUERY TUNING
+#------------------------------------------------------------------------------
+
+# - Planner Method Configuration -
+
+#enable_async_append = on
+#enable_bitmapscan = on
+#enable_gathermerge = on
+#enable_hashagg = on
+#enable_hashjoin = on
+#enable_incremental_sort = on
+#enable_indexscan = on
+#enable_indexonlyscan = on
+#enable_material = on
+#enable_memoize = on
+#enable_mergejoin = on
+#enable_nestloop = on
+#enable_parallel_append = on
+#enable_parallel_hash = on
+#enable_partition_pruning = on
+#enable_partitionwise_join = off
+#enable_partitionwise_aggregate = off
+#enable_presorted_aggregate = on
+#enable_seqscan = on
+#enable_sort = on
+#enable_tidscan = on
+#enable_group_by_reordering = on
+
+# - Planner Cost Constants -
+
+seq_page_cost = 1.0			# 顺序读成本（基准值，保持默认）
+random_page_cost = 2.0			# HDD + RAID 控制器有缓存，比裸 HDD(4.0) 好
+					# 降低此值让优化器更倾向使用索引扫描
+#cpu_tuple_cost = 0.01			# same scale as above
+#cpu_index_tuple_cost = 0.005		# same scale as above
+#cpu_operator_cost = 0.0025		# same scale as above
+#parallel_setup_cost = 1000.0		# same scale as above
+#parallel_tuple_cost = 0.1		# same scale as above
+#min_parallel_table_scan_size = 8MB
+#min_parallel_index_scan_size = 512kB
+effective_cache_size = 94GB		# 服务器 125GB RAM，OS 缓存 + shared_buffers ≈ 75%
+
+#jit_above_cost = 100000		# perform JIT compilation if available
+					# and query more expensive than this;
+					# -1 disables
+#jit_inline_above_cost = 500000		# inline small functions if query is
+					# more expensive than this; -1 disables
+#jit_optimize_above_cost = 500000	# use expensive JIT optimizations if
+					# query is more expensive than this;
+					# -1 disables
+
+# - Genetic Query Optimizer -
+
+#geqo = on
+#geqo_threshold = 12
+#geqo_effort = 5			# range 1-10
+#geqo_pool_size = 0			# selects default based on effort
+#geqo_generations = 0			# selects default based on effort
+#geqo_selection_bias = 2.0		# range 1.5-2.0
+#geqo_seed = 0.0			# range 0.0-1.0
+
+# - Other Planner Options -
+
+#default_statistics_target = 100	# range 1-10000
+#constraint_exclusion = partition	# on, off, or partition
+#cursor_tuple_fraction = 0.1		# range 0.0-1.0
+#from_collapse_limit = 8
+#jit = on				# allow JIT compilation
+#join_collapse_limit = 8		# 1 disables collapsing of explicit
+					# JOIN clauses
+#plan_cache_mode = auto			# auto, force_generic_plan or
+					# force_custom_plan
+#recursive_worktable_factor = 10.0	# range 0.001-1000000
+
+
+#------------------------------------------------------------------------------
+# REPORTING AND LOGGING
+#------------------------------------------------------------------------------
+
+# - Where to Log -
+
+#log_destination = 'stderr'		# Valid values are combinations of
+					# stderr, csvlog, jsonlog, syslog, and
+					# eventlog, depending on platform.
+					# csvlog and jsonlog require
+					# logging_collector to be on.
+
+# This is used when logging to stderr:
+logging_collector = on			# 开启日志收集，记录到文件
+					# (change requires restart)
+
+# These are only used if logging_collector is on:
+#log_directory = 'log'			# directory where log files are written,
+					# can be absolute or relative to PGDATA
+#log_filename = 'postgresql-%Y-%m-%d_%H%M%S.log'	# log file name pattern,
+							# can include strftime() escapes
+#log_file_mode = 0600			# creation mode for log files,
+					# begin with 0 to use octal notation
+#log_rotation_age = 1d			# Automatic rotation of logfiles will
+					# happen after that time.  0 disables.
+#log_rotation_size = 10MB		# Automatic rotation of logfiles will
+					# happen after that much log output.
+					# 0 disables.
+#log_truncate_on_rotation = off		# If on, an existing log file with the
+					# same name as the new log file will be
+					# truncated rather than appended to.
+					# But such truncation only occurs on
+					# time-driven rotation, not on restarts
+					# or size-driven rotation.  Default is
+					# off, meaning append to existing files
+					# in all cases.
+
+# These are relevant when logging to syslog:
+#syslog_facility = 'LOCAL0'
+#syslog_ident = 'postgres'
+#syslog_sequence_numbers = on
+#syslog_split_messages = on
+
+# This is only relevant when logging to eventlog (Windows):
+# (change requires restart)
+#event_source = 'PostgreSQL'
+
+# - When to Log -
+
+#log_min_messages = warning		# values in order of decreasing detail:
+					#   debug5
+					#   debug4
+					#   debug3
+					#   debug2
+					#   debug1
+					#   info
+					#   notice
+					#   warning
+					#   error
+					#   log
+					#   fatal
+					#   panic
+
+#log_min_error_statement = error	# values in order of decreasing detail:
+					#   debug5
+					#   debug4
+					#   debug3
+					#   debug2
+					#   debug1
+					#   info
+					#   notice
+					#   warning
+					#   error
+					#   log
+					#   fatal
+					#   panic (effectively off)
+
+log_min_duration_statement = 500	# 记录超过 500ms 的慢查询，方便排查
+					# -1 禁用, 0 记录全部
+
+#log_min_duration_sample = -1		# -1 is disabled, 0 logs a sample of statements
+					# and their durations, > 0 logs only a sample of
+					# statements running at least this number
+					# of milliseconds;
+					# sample fraction is determined by log_statement_sample_rate
+
+#log_statement_sample_rate = 1.0	# fraction of logged statements exceeding
+					# log_min_duration_sample to be logged;
+					# 1.0 logs all such statements, 0.0 never logs
+
+
+#log_transaction_sample_rate = 0.0	# fraction of transactions whose statements
+					# are logged regardless of their duration; 1.0 logs all
+					# statements from all transactions, 0.0 never logs
+
+#log_startup_progress_interval = 10s	# Time between progress updates for
+					# long-running startup operations.
+					# 0 disables the feature, > 0 indicates
+					# the interval in milliseconds.
+
+# - What to Log -
+
+#debug_print_parse = off
+#debug_print_rewritten = off
+#debug_print_plan = off
+#debug_pretty_print = on
+log_autovacuum_min_duration = 0		# 记录所有 autovacuum 活动，方便监控
+					# -1 disables, 0 logs all actions and
+					# their durations, > 0 logs only
+					# actions running at least this number
+					# of milliseconds.
+#log_checkpoints = on
+#log_connections = off
+#log_disconnections = off
+#log_duration = off
+#log_error_verbosity = default		# terse, default, or verbose messages
+#log_hostname = off
+#log_line_prefix = '%m [%p] '		# special values:
+					#   %a = application name
+					#   %u = user name
+					#   %d = database name
+					#   %r = remote host and port
+					#   %h = remote host
+					#   %b = backend type
+					#   %p = process ID
+					#   %P = process ID of parallel group leader
+					#   %t = timestamp without milliseconds
+					#   %m = timestamp with milliseconds
+					#   %n = timestamp with milliseconds (as a Unix epoch)
+					#   %Q = query ID (0 if none or not computed)
+					#   %i = command tag
+					#   %e = SQL state
+					#   %c = session ID
+					#   %l = session line number
+					#   %s = session start timestamp
+					#   %v = virtual transaction ID
+					#   %x = transaction ID (0 if none)
+					#   %q = stop here in non-session
+					#        processes
+					#   %% = '%'
+					# e.g. '<%u%%%d> '
+#log_lock_waits = off			# log lock waits >= deadlock_timeout
+#log_recovery_conflict_waits = off	# log standby recovery conflict waits
+					# >= deadlock_timeout
+#log_parameter_max_length = -1		# when logging statements, limit logged
+					# bind-parameter values to N bytes;
+					# -1 means print in full, 0 disables
+#log_parameter_max_length_on_error = 0	# when logging an error, limit logged
+					# bind-parameter values to N bytes;
+					# -1 means print in full, 0 disables
+#log_statement = 'none'			# none, ddl, mod, all
+#log_replication_commands = off
+#log_temp_files = -1			# log temporary files equal or larger
+					# than the specified size in kilobytes;
+					# -1 disables, 0 logs all temp files
+log_timezone = 'Asia/Shanghai'
+
+# - Process Title -
+
+#cluster_name = ''			# added to process titles if nonempty
+					# (change requires restart)
+#update_process_title = on
+
+
+#------------------------------------------------------------------------------
+# STATISTICS
+#------------------------------------------------------------------------------
+
+# - Cumulative Query and Index Statistics -
+
+#track_activities = on
+#track_activity_query_size = 1024	# (change requires restart)
+#track_counts = on
+track_io_timing = on			# 记录 I/O 耗时，EXPLAIN (BUFFERS) 可看到
+track_wal_io_timing = on		# 记录 WAL 写入耗时
+#track_functions = none			# none, pl, all
+#stats_fetch_consistency = cache	# cache, none, snapshot
+
+
+# - Monitoring -
+
+#compute_query_id = auto
+#log_statement_stats = off
+#log_parser_stats = off
+#log_planner_stats = off
+#log_executor_stats = off
+
+
+#------------------------------------------------------------------------------
+# AUTOVACUUM
+#------------------------------------------------------------------------------
+
+#autovacuum = on			# Enable autovacuum subprocess?  'on'
+					# requires track_counts to also be on.
+autovacuum_max_workers = 4		# 多个大表可并行 VACUUM
+					# (change requires restart)
+autovacuum_naptime = 30s		# 缩短检查间隔，及时清理死行
+#autovacuum_vacuum_threshold = 50	# min number of row updates before
+					# vacuum
+#autovacuum_vacuum_insert_threshold = 1000	# min number of row inserts
+						# before vacuum; -1 disables insert
+						# vacuums
+#autovacuum_analyze_threshold = 50	# min number of row updates before
+					# analyze
+autovacuum_vacuum_scale_factor = 0.05	# 3000 万行表：5% = 150 万行变更就触发 VACUUM（默认 20% = 600 万）
+autovacuum_vacuum_insert_scale_factor = 0.05	# 同上，插入触发
+autovacuum_analyze_scale_factor = 0.02	# 2% = 60 万行变更就更新统计信息（默认 10% = 300 万）
+#autovacuum_freeze_max_age = 200000000	# maximum XID age before forced vacuum
+					# (change requires restart)
+#autovacuum_multixact_freeze_max_age = 400000000	# maximum multixact age
+							# before forced vacuum
+							# (change requires restart)
+#autovacuum_vacuum_cost_delay = 2ms	# default vacuum cost delay for
+					# autovacuum, in milliseconds;
+					# -1 means use vacuum_cost_delay
+#autovacuum_vacuum_cost_limit = -1	# default vacuum cost limit for
+					# autovacuum, -1 means use
+					# vacuum_cost_limit
+
+
+#------------------------------------------------------------------------------
+# CLIENT CONNECTION DEFAULTS
+#------------------------------------------------------------------------------
+
+# - Statement Behavior -
+
+#client_min_messages = notice		# values in order of decreasing detail:
+					#   debug5
+					#   debug4
+					#   debug3
+					#   debug2
+					#   debug1
+					#   log
+					#   notice
+					#   warning
+					#   error
+#search_path = '"$user", public'	# schema names
+#row_security = on
+#default_table_access_method = 'heap'
+#default_tablespace = ''		# a tablespace name, '' uses the default
+#default_toast_compression = 'pglz'	# 'pglz' or 'lz4'
+#temp_tablespaces = ''			# a list of tablespace names, '' uses
+					# only default tablespace
+#check_function_bodies = on
+#default_transaction_isolation = 'read committed'
+#default_transaction_read_only = off
+#default_transaction_deferrable = off
+#session_replication_role = 'origin'
+#statement_timeout = 0				# in milliseconds, 0 is disabled
+#transaction_timeout = 0			# in milliseconds, 0 is disabled
+#lock_timeout = 0				# in milliseconds, 0 is disabled
+#idle_in_transaction_session_timeout = 0	# in milliseconds, 0 is disabled
+#idle_session_timeout = 0			# in milliseconds, 0 is disabled
+#vacuum_freeze_table_age = 150000000
+#vacuum_freeze_min_age = 50000000
+#vacuum_failsafe_age = 1600000000
+#vacuum_multixact_freeze_table_age = 150000000
+#vacuum_multixact_freeze_min_age = 5000000
+#vacuum_multixact_failsafe_age = 1600000000
+#bytea_output = 'hex'			# hex, escape
+#xmlbinary = 'base64'
+#xmloption = 'content'
+#gin_pending_list_limit = 4MB
+#createrole_self_grant = ''		# set and/or inherit
+#event_triggers = on
+
+# - Locale and Formatting -
+
+datestyle = 'iso, mdy'
+#intervalstyle = 'postgres'
+timezone = 'Asia/Shanghai'
+#timezone_abbreviations = 'Default'	# Select the set of available time zone
+					# abbreviations.  Currently, there are
+					#   Default
+					#   Australia (historical usage)
+					#   India
+					# You can create your own file in
+					# share/timezonesets/.
+#extra_float_digits = 1			# min -15, max 3; any value >0 actually
+					# selects precise output mode
+#client_encoding = sql_ascii		# actually, defaults to database
+					# encoding
+
+# These settings are initialized by initdb, but they can be changed.
+lc_messages = 'en_US.utf8'		# locale for system error message
+					# strings
+lc_monetary = 'en_US.utf8'		# locale for monetary formatting
+lc_numeric = 'en_US.utf8'		# locale for number formatting
+lc_time = 'en_US.utf8'			# locale for time formatting
+
+#icu_validation_level = warning		# report ICU locale validation
+					# errors at the given level
+
+# default configuration for text search
+default_text_search_config = 'pg_catalog.english'
+
+# - Shared Library Preloading -
+
+#local_preload_libraries = ''
+#session_preload_libraries = ''
+shared_preload_libraries = 'pg_stat_statements'	# 启用慢查询统计（change requires restart）
+#jit_provider = 'llvmjit'		# JIT library to use
+
+# - Other Defaults -
+
+#dynamic_library_path = '$libdir'
+#extension_destdir = ''			# prepend path when loading extensions
+					# and shared objects (added by Debian)
+#gin_fuzzy_search_limit = 0
+
+
+#------------------------------------------------------------------------------
+# LOCK MANAGEMENT
+#------------------------------------------------------------------------------
+
+#deadlock_timeout = 1s
+#max_locks_per_transaction = 64		# min 10
+					# (change requires restart)
+#max_pred_locks_per_transaction = 64	# min 10
+					# (change requires restart)
+#max_pred_locks_per_relation = -2	# negative values mean
+					# (max_pred_locks_per_transaction
+					#  / -max_pred_locks_per_relation) - 1
+#max_pred_locks_per_page = 2		# min 0
+
+
+#------------------------------------------------------------------------------
+# VERSION AND PLATFORM COMPATIBILITY
+#------------------------------------------------------------------------------
+
+# - Previous PostgreSQL Versions -
+
+#array_nulls = on
+#backslash_quote = safe_encoding	# on, off, or safe_encoding
+#escape_string_warning = on
+#lo_compat_privileges = off
+#quote_all_identifiers = off
+#standard_conforming_strings = on
+#synchronize_seqscans = on
+
+# - Other Platforms and Clients -
+
+#transform_null_equals = off
+#allow_alter_system = on
+
+
+#------------------------------------------------------------------------------
+# ERROR HANDLING
+#------------------------------------------------------------------------------
+
+#exit_on_error = off			# terminate session on any error?
+#restart_after_crash = on		# reinitialize after backend crash?
+#data_sync_retry = off			# retry or panic on failure to fsync
+					# data?
+					# (change requires restart)
+#recovery_init_sync_method = fsync	# fsync, syncfs (Linux 5.8+)
+
+
+#------------------------------------------------------------------------------
+# CONFIG FILE INCLUDES
+#------------------------------------------------------------------------------
+
+# These options allow settings to be loaded from files other than the
+# default postgresql.conf.  Note that these are directives, not variable
+# assignments, so they can usefully be given more than once.
+
+#include_dir = '...'			# include files ending in '.conf' from
+					# a directory, e.g., 'conf.d'
+#include_if_exists = '...'		# include file only if it exists
+#include = '...'			# include file
+
+
+#------------------------------------------------------------------------------
+# CUSTOMIZED OPTIONS
+#------------------------------------------------------------------------------
+
+# Add settings for extensions here
--- a/scripts/perf_query/seed.go
+++ b/scripts/perf_query/seed.go
@@ -0,0 +1,700 @@
+//go:build ignore
+
+// 性能测试数据生成/清理脚本
+// 用法:
+//   source .env.local
+//   go run ./scripts/perf_query/seed.go                       # 默认生成 3000 万卡
+//   go run ./scripts/perf_query/seed.go -total 1000000        # 生成 100 万卡（试跑）
+//   go run ./scripts/perf_query/seed.go -action cleanup       # 清理测试数据
+//   go run ./scripts/perf_query/seed.go -action verify        # 验证数据分布
+//   go run ./scripts/perf_query/seed.go -action add-index     # 创建优化索引
+//   go run ./scripts/perf_query/seed.go -action drop-index    # 删除优化索引
+
+package main
+
+import (
+	"database/sql"
+	"flag"
+	"fmt"
+	"log"
+	"os"
+	"strconv"
+	"strings"
+	"sync"
+	"sync/atomic"
+	"time"
+
+	_ "github.com/jackc/pgx/v5/stdlib"
+	"golang.org/x/crypto/bcrypt"
+	"gorm.io/driver/postgres"
+	"gorm.io/gorm"
+	"gorm.io/gorm/logger"
+)
+
+// 测试数据标记常量
+const (
+	testBatchNo        = "PERF-TEST-30M"
+	testShopCodePrefix = "PERF-TEST-SHOP-"
+	testAgentUsername  = "perf_test_agent"
+	testAgentPassword  = "PerfTest@123456"
+	numShops           = 200
+	agentChildShops    = 6 // 代理商有 6 个下级店铺（自己 + 6 = 7 个总共）
+)
+
+// 简化模型（仅用于创建测试数据，不引入项目内部包）
+type testShop struct {
+	ID           uint   `gorm:"primaryKey"`
+	ShopName     string `gorm:"column:shop_name"`
+	ShopCode     string `gorm:"column:shop_code"`
+	ParentID     *uint  `gorm:"column:parent_id"`
+	Level        int    `gorm:"column:level"`
+	ContactName  string `gorm:"column:contact_name"`
+	ContactPhone string `gorm:"column:contact_phone"`
+	Status       int    `gorm:"column:status"`
+	Creator      uint   `gorm:"column:creator"`
+	Updater      uint   `gorm:"column:updater"`
+	CreatedAt    time.Time
+	UpdatedAt    time.Time
+}
+
+func (testShop) TableName() string { return "tb_shop" }
+
+type testAccount struct {
+	ID           uint   `gorm:"primaryKey"`
+	Username     string `gorm:"column:username"`
+	Phone        string `gorm:"column:phone"`
+	Password     string `gorm:"column:password"`
+	UserType     int    `gorm:"column:user_type"`
+	ShopID       *uint  `gorm:"column:shop_id"`
+	EnterpriseID *uint  `gorm:"column:enterprise_id"`
+	Status       int    `gorm:"column:status"`
+	Creator      uint   `gorm:"column:creator"`
+	Updater      uint   `gorm:"column:updater"`
+	CreatedAt    time.Time
+	UpdatedAt    time.Time
+}
+
+func (testAccount) TableName() string { return "tb_account" }
+
+var (
+	action    = flag.String("action", "seed", "操作: seed / cleanup / verify / add-index / drop-index")
+	total     = flag.Int("total", 30000000, "生成卡总数")
+	batchSize = flag.Int("batch", 5000000, "每批插入数量")
+)
+
+func buildDSN() string {
+	return fmt.Sprintf("host=%s port=%s user=%s password=%s dbname=%s sslmode=disable TimeZone=Asia/Shanghai",
+		getEnv("JUNHONG_DATABASE_HOST", "cxd.whcxd.cn"),
+		getEnv("JUNHONG_DATABASE_PORT", "16159"),
+		getEnv("JUNHONG_DATABASE_USER", "erp_pgsql"),
+		getEnv("JUNHONG_DATABASE_PASSWORD", "erp_2025"),
+		getEnv("JUNHONG_DATABASE_DBNAME", "junhong_cmp_test"),
+	)
+}
+
+func main() {
+	flag.Parse()
+
+	db, err := gorm.Open(postgres.Open(buildDSN()), &gorm.Config{
+		Logger: logger.Default.LogMode(logger.Warn),
+	})
+	if err != nil {
+		log.Fatalf("❌ 连接数据库失败: %v", err)
+	}
+
+	sqlDB, _ := db.DB()
+	sqlDB.SetMaxOpenConns(5)
+	sqlDB.SetMaxIdleConns(3)
+	fmt.Println("✅ 数据库连接成功")
+
+	switch *action {
+	case "seed":
+		doSeed(db)
+	case "cleanup":
+		doCleanup(db)
+	case "verify":
+		doVerify(db)
+	case "add-index":
+		doAddIndex(db)
+	case "drop-index":
+		doDropIndex(db)
+	default:
+		log.Fatalf("❌ 未知操作: %s（支持: seed / cleanup / verify / add-index / drop-index）", *action)
+	}
+}
+
+// ==================== Seed ====================
+
+func doSeed(db *gorm.DB) {
+	fmt.Println("\n🚀 开始生成性能测试数据（UNLOGGED 极速模式）")
+	fmt.Printf("   目标卡数: %d\n", *total)
+	fmt.Printf("   店铺数量: %d（代理商可见 %d 个）\n", numShops, agentChildShops+1)
+
+	overallStart := time.Now()
+
+	fmt.Println("\n📦 步骤 1/8: 创建测试店铺...")
+	shopIDs := createTestShops(db)
+	fmt.Printf("   ✅ 创建了 %d 个店铺\n", len(shopIDs))
+
+	fmt.Println("\n👤 步骤 2/8: 创建测试代理商账号...")
+	createTestAgent(db, shopIDs[0])
+	fmt.Printf("   ✅ 账号: %s / %s (shop_id=%d)\n", testAgentUsername, testAgentPassword, shopIDs[0])
+
+	// 调高 PG 配置减少 checkpoint 频率
+	fmt.Println("\n⚙️  步骤 3/8: 临时调高 PostgreSQL 写入配置...")
+	tuneForBulkLoad(db)
+
+	// 先删索引再插入，速度提升 10~50 倍
+	fmt.Println("\n📇 步骤 4/8: 临时删除非主键索引（加速批量插入）...")
+	droppedIndexes := dropNonPKIndexes(db)
+	fmt.Printf("   ✅ 删除了 %d 个索引\n", len(droppedIndexes))
+
+	// 设置 UNLOGGED 跳过 WAL 写入（表数据少时几乎瞬间完成）
+	fmt.Println("\n⚡ 步骤 5/8: 设置表为 UNLOGGED（跳过 WAL，极速写入）...")
+	start := time.Now()
+	if err := db.Exec("ALTER TABLE tb_iot_card SET UNLOGGED").Error; err != nil {
+		log.Fatalf("设置 UNLOGGED 失败: %v", err)
+	}
+	fmt.Printf("   ✅ 已设置 UNLOGGED (耗时 %v)\n", time.Since(start))
+
+	fmt.Println("\n💳 步骤 6/8: 批量插入卡数据（UNLOGGED + generate_series）...")
+	insertCards(db, shopIDs)
+
+	// 恢复为 LOGGED（会重写整张表到 WAL，30M 行约需 3-5 分钟）
+	fmt.Println("\n🔒 步骤 7/8: 恢复表为 LOGGED（重写 WAL，请耐心等待）...")
+	start = time.Now()
+	if err := db.Exec("ALTER TABLE tb_iot_card SET LOGGED").Error; err != nil {
+		// SET LOGGED 失败不致命，表仍然可用，只是崩溃会丢数据
+		fmt.Printf("   ⚠️  恢复 LOGGED 失败（测试数据可接受）: %v\n", err)
+	} else {
+		fmt.Printf("   ✅ 已恢复 LOGGED (耗时 %v)\n", time.Since(start))
+	}
+
+	fmt.Println("\n📇 步骤 8/8: 重建索引 + ANALYZE...")
+	rebuildIndexes(db, droppedIndexes)
+	fmt.Printf("   ✅ 重建了 %d 个索引\n", len(droppedIndexes))
+
+	start = time.Now()
+	db.Exec("ANALYZE tb_iot_card")
+	fmt.Printf("   ✅ ANALYZE 完成 (耗时 %v)\n", time.Since(start))
+
+	// 恢复 PG 配置
+	restoreAfterBulkLoad(db)
+
+	fmt.Printf("\n🎉 全部完成！总耗时: %v\n", time.Since(overallStart))
+	fmt.Println("\n📝 后续步骤:")
+	fmt.Println("   1. 启动 API 服务: source .env.local && go run ./cmd/api/...")
+	fmt.Printf("   2. 运行基准测试: go run ./scripts/perf_query/bench.go\n")
+	fmt.Println("   3. 测试完成后清理: go run ./scripts/perf_query/seed.go -action cleanup")
+}
+
+// tuneForBulkLoad 临时调高 PG 配置以加速批量写入
+func tuneForBulkLoad(db *gorm.DB) {
+	settings := []struct {
+		sql  string
+		desc string
+	}{
+		{"ALTER SYSTEM SET max_wal_size = '4GB'", "max_wal_size → 4GB（减少 checkpoint 频率）"},
+		{"ALTER SYSTEM SET checkpoint_timeout = '30min'", "checkpoint_timeout → 30min"},
+		{"ALTER SYSTEM SET synchronous_commit = 'off'", "synchronous_commit → off（不等待 WAL 刷盘）"},
+	}
+	for _, s := range settings {
+		if err := db.Exec(s.sql).Error; err != nil {
+			fmt.Printf("   ⚠️  %s 失败: %v\n", s.desc, err)
+		} else {
+			fmt.Printf("   ✅ %s\n", s.desc)
+		}
+	}
+	db.Exec("SELECT pg_reload_conf()")
+	fmt.Println("   ✅ 配置已重载")
+}
+
+// restoreAfterBulkLoad 恢复 PG 配置
+func restoreAfterBulkLoad(db *gorm.DB) {
+	fmt.Println("\n⚙️  恢复 PostgreSQL 配置...")
+	db.Exec("ALTER SYSTEM SET max_wal_size = '1GB'")
+	db.Exec("ALTER SYSTEM SET checkpoint_timeout = '5min'")
+	db.Exec("ALTER SYSTEM SET synchronous_commit = 'on'")
+	db.Exec("SELECT pg_reload_conf()")
+	fmt.Println("   ✅ PG 配置已恢复默认")
+}
+
+func createTestShops(db *gorm.DB) []uint {
+	now := time.Now()
+	shopIDs := make([]uint, 0, numShops)
+
+	// 先清理可能存在的旧测试店铺
+	db.Exec("DELETE FROM tb_shop WHERE shop_code LIKE ?", testShopCodePrefix+"%")
+
+	// 创建根店铺（代理商总店）
+	rootShop := testShop{
+		ShopName:     "性能测试-总代理",
+		ShopCode:     testShopCodePrefix + "ROOT",
+		Level:        1,
+		ContactName:  "测试联系人",
+		ContactPhone: "13800000000",
+		Status:       1,
+		Creator:      1,
+		Updater:      1,
+		CreatedAt:    now,
+		UpdatedAt:    now,
+	}
+	if err := db.Create(&rootShop).Error; err != nil {
+		log.Fatalf("创建根店铺失败: %v", err)
+	}
+	shopIDs = append(shopIDs, rootShop.ID)
+
+	// 创建 6 个子店铺（代理商下级）
+	for i := 1; i <= agentChildShops; i++ {
+		child := testShop{
+			ShopName:     fmt.Sprintf("性能测试-分店%d", i),
+			ShopCode:     fmt.Sprintf("%sSUB-%d", testShopCodePrefix, i),
+			ParentID:     &rootShop.ID,
+			Level:        2,
+			ContactName:  "测试联系人",
+			ContactPhone: fmt.Sprintf("1380000%04d", i),
+			Status:       1,
+			Creator:      1,
+			Updater:      1,
+			CreatedAt:    now,
+			UpdatedAt:    now,
+		}
+		if err := db.Create(&child).Error; err != nil {
+			log.Fatalf("创建子店铺失败: %v", err)
+		}
+		shopIDs = append(shopIDs, child.ID)
+	}
+
+	// 创建其余独立店铺（其他代理商）
+	for i := agentChildShops + 1; i < numShops; i++ {
+		indep := testShop{
+			ShopName:     fmt.Sprintf("性能测试-独立店铺%d", i),
+			ShopCode:     fmt.Sprintf("%sINDEP-%d", testShopCodePrefix, i),
+			Level:        1,
+			ContactName:  "测试联系人",
+			ContactPhone: fmt.Sprintf("1390000%04d", i),
+			Status:       1,
+			Creator:      1,
+			Updater:      1,
+			CreatedAt:    now,
+			UpdatedAt:    now,
+		}
+		if err := db.Create(&indep).Error; err != nil {
+			log.Fatalf("创建独立店铺失败: %v", err)
+		}
+		shopIDs = append(shopIDs, indep.ID)
+	}
+
+	return shopIDs
+}
+
+func createTestAgent(db *gorm.DB, shopID uint) {
+	// 先清理可能存在的旧测试账号
+	db.Exec("DELETE FROM tb_account WHERE username = ?", testAgentUsername)
+
+	// bcrypt 加密密码
+	hashedPassword, err := bcrypt.GenerateFromPassword([]byte(testAgentPassword), bcrypt.DefaultCost)
+	if err != nil {
+		log.Fatalf("加密密码失败: %v", err)
+	}
+
+	account := testAccount{
+		Username:  testAgentUsername,
+		Phone:     "19999999999",
+		Password:  string(hashedPassword),
+		UserType:  3, // 代理账号
+		ShopID:    &shopID,
+		Status:    1,
+		Creator:   1,
+		Updater:   1,
+		CreatedAt: time.Now(),
+		UpdatedAt: time.Now(),
+	}
+	if err := db.Create(&account).Error; err != nil {
+		log.Fatalf("创建代理账号失败: %v", err)
+	}
+}
+
+func insertCards(db *gorm.DB, shopIDs []uint) {
+	idStrs := make([]string, len(shopIDs))
+	for i, id := range shopIDs {
+		idStrs[i] = strconv.FormatUint(uint64(id), 10)
+	}
+	pgArray := "ARRAY[" + strings.Join(idStrs, ",") + "]::bigint[]"
+
+	sqlTemplate := fmt.Sprintf(`
+INSERT INTO tb_iot_card (
+    iccid, card_category, carrier_id, carrier_type, carrier_name,
+    imsi, msisdn, batch_no, supplier, cost_price, distribute_price,
+    status, shop_id, activation_status, real_name_status, network_status,
+    data_usage_mb, enable_polling, created_at, updated_at, creator, updater, series_id,
+    is_standalone
+)
+SELECT
+    'T' || lpad(i::text, 19, '0'),
+    CASE WHEN random() < 0.85 THEN 'normal' ELSE 'industry' END,
+    (i %% 4 + 1),
+    CASE (i %% 4) WHEN 0 THEN 'CMCC' WHEN 1 THEN 'CUCC' WHEN 2 THEN 'CTCC' ELSE 'CBN' END,
+    CASE (i %% 4) WHEN 0 THEN '中国移动' WHEN 1 THEN '中国联通' WHEN 2 THEN '中国电信' ELSE '中国广电' END,
+    lpad(i::text, 15, '0'),
+    '1' || lpad((i %% 10000000)::text, 10, '0'),
+    '%s',
+    CASE (i %% 3) WHEN 0 THEN '供应商A' WHEN 1 THEN '供应商B' ELSE '供应商C' END,
+    (random() * 5000 + 500)::bigint,
+    (random() * 3000 + 1000)::bigint,
+    CASE WHEN random() < 0.15 THEN 1 WHEN random() < 0.30 THEN 2 WHEN random() < 0.90 THEN 3 ELSE 4 END,
+    (%s)[GREATEST(1, LEAST(%d, floor(%d * power(random(), 2.5))::int + 1))],
+    CASE WHEN random() < 0.7 THEN 1 ELSE 0 END,
+    CASE WHEN random() < 0.6 THEN 1 ELSE 0 END,
+    CASE WHEN random() < 0.8 THEN 1 ELSE 0 END,
+    (random() * 100000)::bigint,
+    random() < 0.9,
+    now() - interval '1 day' * (random() * 730),
+    now() - interval '1 day' * (random() * 30),
+    1, 1,
+    CASE WHEN random() < 0.8 THEN (random() * 20 + 1)::int ELSE NULL END,
+    true
+FROM generate_series($1::bigint, $2::bigint) AS s(i)
+`, testBatchNo, pgArray, numShops, numShops)
+
+	// 500 万/批 × 4 并发 worker（HDD 上减少 I/O 竞争）
+	const workerCount = 4
+	const chunkSize = 5000000
+
+	type task struct {
+		batchNum int
+		start    int
+		end      int
+	}
+
+	totalBatches := (*total + chunkSize - 1) / chunkSize
+	taskCh := make(chan task, totalBatches)
+	var wg sync.WaitGroup
+	var insertedCount int64
+	var errCount int64
+	totalStart := time.Now()
+
+	dsn := buildDSN()
+
+	for w := 0; w < workerCount; w++ {
+		wg.Add(1)
+		go func(workerID int) {
+			defer wg.Done()
+			conn, err := sql.Open("pgx", dsn)
+			if err != nil {
+				log.Printf("Worker %d 连接失败: %v", workerID, err)
+				return
+			}
+			defer conn.Close()
+			conn.SetMaxOpenConns(1)
+			conn.Exec("SET synchronous_commit = off")
+			conn.Exec("SET work_mem = '256MB'")
+
+			for t := range taskCh {
+				_, err := conn.Exec(sqlTemplate, t.start, t.end)
+				if err != nil {
+					log.Printf("Worker %d 批次 %d 失败: %v", workerID, t.batchNum, err)
+					atomic.AddInt64(&errCount, 1)
+					continue
+				}
+
+				count := t.end - t.start + 1
+				done := atomic.AddInt64(&insertedCount, int64(count))
+				elapsed := time.Since(totalStart).Seconds()
+				speed := float64(done) / elapsed
+				remaining := float64(int64(*total)-done) / speed
+				fmt.Printf("   [W%d] 批次 %d/%d 完成 | 累计 %d/%d (%.1f%%) | %.0f条/秒 | 剩余 %.0f秒\n",
+					workerID, t.batchNum, totalBatches, done, *total,
+					float64(done)*100/float64(*total), speed, remaining)
+			}
+		}(w)
+	}
+
+	for b := 0; b < totalBatches; b++ {
+		start := b*chunkSize + 1
+		end := (b + 1) * chunkSize
+		if end > *total {
+			end = *total
+		}
+		taskCh <- task{batchNum: b + 1, start: start, end: end}
+	}
+	close(taskCh)
+	wg.Wait()
+
+	elapsed := time.Since(totalStart)
+	fmt.Printf("   ✅ 共插入 %d 条 (失败 %d 批) | 总耗时 %v | 平均 %.0f 条/秒\n",
+		atomic.LoadInt64(&insertedCount), atomic.LoadInt64(&errCount), elapsed.Round(time.Second),
+		float64(atomic.LoadInt64(&insertedCount))/elapsed.Seconds())
+}
+
+// ==================== Cleanup ====================
+
+func doCleanup(db *gorm.DB) {
+	fmt.Println("\n🧹 开始清理性能测试数据")
+
+	// 删除测试卡数据（分批删除）
+	fmt.Println("\n💳 删除测试卡数据...")
+	totalDeleted := 0
+	start := time.Now()
+	for {
+		result := db.Exec("DELETE FROM tb_iot_card WHERE ctid IN (SELECT ctid FROM tb_iot_card WHERE batch_no = ? LIMIT 500000)", testBatchNo)
+		if result.Error != nil {
+			log.Fatalf("删除失败: %v", result.Error)
+		}
+		if result.RowsAffected == 0 {
+			break
+		}
+		totalDeleted += int(result.RowsAffected)
+		elapsed := time.Since(start)
+		speed := float64(totalDeleted) / elapsed.Seconds()
+		fmt.Printf("   已删除 %d 条 (%.0f 条/秒)\n", totalDeleted, speed)
+	}
+	fmt.Printf("   ✅ 共删除 %d 条卡数据 (耗时 %v)\n", totalDeleted, time.Since(start).Round(time.Second))
+
+	// 删除测试店铺
+	fmt.Println("\n📦 删除测试店铺...")
+	result := db.Exec("DELETE FROM tb_shop WHERE shop_code LIKE ?", testShopCodePrefix+"%")
+	fmt.Printf("   ✅ 删除了 %d 个店铺\n", result.RowsAffected)
+
+	// 删除测试账号
+	fmt.Println("\n👤 删除测试账号...")
+	result = db.Exec("DELETE FROM tb_account WHERE username = ?", testAgentUsername)
+	fmt.Printf("   ✅ 删除了 %d 个账号\n", result.RowsAffected)
+
+	// 删除优化索引（如果存在）
+	fmt.Println("\n📇 清理优化索引...")
+	db.Exec("DROP INDEX IF EXISTS idx_iot_card_perf_shop_created")
+	db.Exec("DROP INDEX IF EXISTS idx_iot_card_perf_shop_status_created")
+	fmt.Println("   ✅ 索引已清理")
+
+	fmt.Println("\n⚠️  建议手动执行 VACUUM FULL tb_iot_card; 回收磁盘空间")
+	fmt.Println("🎉 清理完成！")
+}
+
+// ==================== Verify ====================
+
+func doVerify(db *gorm.DB) {
+	fmt.Println("\n📊 验证测试数据分布")
+
+	// 总卡数
+	var total int64
+	db.Raw("SELECT count(*) FROM tb_iot_card WHERE batch_no = ? AND deleted_at IS NULL", testBatchNo).Scan(&total)
+	fmt.Printf("\n总测试卡数: %d\n", total)
+
+	// 代理商可见卡数
+	var agentVisible int64
+	db.Raw(`
+		SELECT count(*) FROM tb_iot_card 
+		WHERE batch_no = ? AND deleted_at IS NULL 
+		AND shop_id IN (SELECT id FROM tb_shop WHERE shop_code LIKE ? AND deleted_at IS NULL)
+	`, testBatchNo, testShopCodePrefix+"ROOT").Scan(&agentVisible)
+
+	var agentSubVisible int64
+	db.Raw(`
+		SELECT count(*) FROM tb_iot_card 
+		WHERE batch_no = ? AND deleted_at IS NULL 
+		AND shop_id IN (
+			SELECT id FROM tb_shop 
+			WHERE (shop_code LIKE ? OR shop_code LIKE ?) AND deleted_at IS NULL
+		)
+	`, testBatchNo, testShopCodePrefix+"ROOT", testShopCodePrefix+"SUB-%").Scan(&agentSubVisible)
+	fmt.Printf("代理商总店卡数: %d\n", agentVisible)
+	fmt.Printf("代理商(含下级)卡数: %d (占比 %.1f%%)\n", agentSubVisible, float64(agentSubVisible)*100/float64(total))
+
+	// 前 20 名店铺分布
+	type shopDist struct {
+		ShopID uint
+		Cnt    int64
+	}
+	var dists []shopDist
+	db.Raw(`
+		SELECT shop_id, count(*) as cnt FROM tb_iot_card 
+		WHERE batch_no = ? AND deleted_at IS NULL 
+		GROUP BY shop_id ORDER BY cnt DESC LIMIT 20
+	`, testBatchNo).Scan(&dists)
+
+	fmt.Println("\n前 20 名店铺卡量分布:")
+	fmt.Printf("  %-12s %-12s %-8s\n", "shop_id", "卡数", "占比")
+	for _, d := range dists {
+		fmt.Printf("  %-12d %-12d %.2f%%\n", d.ShopID, d.Cnt, float64(d.Cnt)*100/float64(total))
+	}
+
+	// 状态分布
+	type statusDist struct {
+		Status int
+		Cnt    int64
+	}
+	var statuses []statusDist
+	db.Raw("SELECT status, count(*) as cnt FROM tb_iot_card WHERE batch_no = ? AND deleted_at IS NULL GROUP BY status ORDER BY status", testBatchNo).Scan(&statuses)
+
+	fmt.Println("\n状态分布:")
+	statusNames := map[int]string{1: "在库", 2: "已分销", 3: "已激活", 4: "已停用"}
+	for _, s := range statuses {
+		fmt.Printf("  状态 %d (%s): %d (%.1f%%)\n", s.Status, statusNames[s.Status], s.Cnt, float64(s.Cnt)*100/float64(total))
+	}
+
+	// 运营商分布
+	type carrierDist struct {
+		CarrierID uint
+		Cnt       int64
+	}
+	var carriers []carrierDist
+	db.Raw("SELECT carrier_id, count(*) as cnt FROM tb_iot_card WHERE batch_no = ? AND deleted_at IS NULL GROUP BY carrier_id ORDER BY carrier_id", testBatchNo).Scan(&carriers)
+
+	fmt.Println("\n运营商分布:")
+	for _, c := range carriers {
+		fmt.Printf("  carrier_id=%d: %d (%.1f%%)\n", c.CarrierID, c.Cnt, float64(c.Cnt)*100/float64(total))
+	}
+
+	// 表大小
+	var tableSize, indexSize string
+	db.Raw("SELECT pg_size_pretty(pg_relation_size('tb_iot_card'))").Scan(&tableSize)
+	db.Raw("SELECT pg_size_pretty(pg_indexes_size('tb_iot_card'))").Scan(&indexSize)
+	fmt.Printf("\n表大小: %s | 索引大小: %s\n", tableSize, indexSize)
+
+	// 测试代理商账号信息
+	var agentInfo struct {
+		ID       uint
+		Username string
+		ShopID   uint
+	}
+	db.Raw("SELECT id, username, shop_id FROM tb_account WHERE username = ? AND deleted_at IS NULL", testAgentUsername).Scan(&agentInfo)
+	fmt.Printf("\n测试代理商: id=%d, username=%s, shop_id=%d\n", agentInfo.ID, agentInfo.Username, agentInfo.ShopID)
+}
+
+// ==================== Bulk Load Index Management ====================
+
+type indexDef struct {
+	Name       string
+	Definition string
+}
+
+func dropNonPKIndexes(db *gorm.DB) []indexDef {
+	sqlDB, _ := db.DB()
+
+	var indexes []indexDef
+	rows, err := sqlDB.Query(`
+		SELECT indexname, indexdef FROM pg_indexes 
+		WHERE tablename = 'tb_iot_card' AND indexname != 'iot_cards_pkey'
+		ORDER BY indexname
+	`)
+	if err != nil {
+		log.Fatalf("查询索引失败: %v", err)
+	}
+	defer rows.Close()
+
+	for rows.Next() {
+		var idx indexDef
+		rows.Scan(&idx.Name, &idx.Definition)
+		indexes = append(indexes, idx)
+	}
+
+	for _, idx := range indexes {
+		fmt.Printf("   删除 %s ...\n", idx.Name)
+		sqlDB.Exec("DROP INDEX IF EXISTS " + idx.Name)
+	}
+
+	return indexes
+}
+
+func rebuildIndexes(db *gorm.DB, indexes []indexDef) {
+	sqlDB, _ := db.DB()
+
+	for _, idx := range indexes {
+		fmt.Printf("   重建 %s ...", idx.Name)
+		start := time.Now()
+		if _, err := sqlDB.Exec(idx.Definition); err != nil {
+			fmt.Printf(" ❌ 失败: %v\n", err)
+			continue
+		}
+		fmt.Printf(" ✅ (%v)\n", time.Since(start).Round(time.Second))
+	}
+}
+
+// ==================== Optimization Index Management ====================
+
+func doAddIndex(db *gorm.DB) {
+	fmt.Println("\n📇 创建优化复合索引（CONCURRENTLY，不阻塞读写）")
+
+	indexes := []struct {
+		name string
+		sql  string
+	}{
+		{
+			name: "idx_iot_card_perf_shop_created",
+			sql:  "CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_iot_card_perf_shop_created ON tb_iot_card (shop_id, created_at DESC) WHERE deleted_at IS NULL",
+		},
+		{
+			name: "idx_iot_card_perf_shop_status_created",
+			sql:  "CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_iot_card_perf_shop_status_created ON tb_iot_card (shop_id, status, created_at DESC) WHERE deleted_at IS NULL",
+		},
+	}
+
+	// CONCURRENTLY 不能在事务内执行，直接用底层连接
+	sqlDB, _ := db.DB()
+
+	// pg_trgm GIN 索引：加速 ICCID/MSISDN 中间模糊搜索（LIKE '%xxx%'）
+	fmt.Println("\n   启用 pg_trgm 扩展...")
+	if _, err := sqlDB.Exec("CREATE EXTENSION IF NOT EXISTS pg_trgm"); err != nil {
+		fmt.Printf("   ⚠️  pg_trgm 扩展创建失败（可能需要超级管理员权限）: %v\n", err)
+	} else {
+		fmt.Println("   ✅ pg_trgm 扩展已启用")
+		trgmIndexes := []struct {
+			name string
+			sql  string
+		}{
+			{
+				name: "idx_iot_card_iccid_trgm",
+				sql:  "CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_iot_card_iccid_trgm ON tb_iot_card USING gin (iccid gin_trgm_ops)",
+			},
+			{
+				name: "idx_iot_card_msisdn_trgm",
+				sql:  "CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_iot_card_msisdn_trgm ON tb_iot_card USING gin (msisdn gin_trgm_ops)",
+			},
+		}
+		for _, idx := range trgmIndexes {
+			indexes = append(indexes, struct {
+				name string
+				sql  string
+			}{idx.name, idx.sql})
+		}
+	}
+
+	for _, idx := range indexes {
+		fmt.Printf("   创建 %s ...", idx.name)
+		start := time.Now()
+		if _, err := sqlDB.Exec(idx.sql); err != nil {
+			fmt.Printf(" ❌ 失败: %v\n", err)
+			continue
+		}
+		fmt.Printf(" ✅ (%v)\n", time.Since(start).Round(time.Second))
+	}
+
+	// 删除重复索引
+	fmt.Println("   清理重复索引 idx_tb_iot_card_shop_id（与 idx_iot_card_shop_id 重复）...")
+	sqlDB.Exec("DROP INDEX IF EXISTS idx_tb_iot_card_shop_id")
+
+	fmt.Println("\n   更新统计信息...")
+	sqlDB.Exec("ANALYZE tb_iot_card")
+	fmt.Println("   ✅ 索引优化完成")
+}
+
+func doDropIndex(db *gorm.DB) {
+	fmt.Println("\n📇 删除优化索引（恢复到原始状态）")
+	sqlDB, _ := db.DB()
+	sqlDB.Exec("DROP INDEX IF EXISTS idx_iot_card_perf_shop_created")
+	sqlDB.Exec("DROP INDEX IF EXISTS idx_iot_card_perf_shop_status_created")
+	sqlDB.Exec("DROP INDEX IF EXISTS idx_iot_card_iccid_trgm")
+	sqlDB.Exec("DROP INDEX IF EXISTS idx_iot_card_msisdn_trgm")
+	sqlDB.Exec("ANALYZE tb_iot_card")
+	fmt.Println("   ✅ 已删除优化索引")
+}
+
+// ==================== Utils ====================
+
+func getEnv(key, fallback string) string {
+	if v := os.Getenv(key); v != "" {
+		return v
+	}
+	return fallback
+}